My last blog talked about Dell acquiring EqualLogic.  I received a message from a LeftHand end user customer with a number of good questions.  It made me think of what the Dell acquisition of EqualLogic has on the storage and iSCSI market. 

- Dell buying EqualLogic should at first be a rising tide for all iSCSI players.  It raises awareness and further validates the market.  End users want to mitigate risk - and if Dell is willing to invest in iSCSI at the tune of $1.4 B - then the end user should feel more comfortable about making an investment as well.

- All of the major storage vendors will step up their investments in iSCSI.  You will see more targeted sales efforts and marketing around everyone's iSCSI storage systems.  Again, this will further raise awareness around iSCSI and the tide will rise even further for one and all. 

- End users of the other emerging iSCSI vendors may be questioning their decision - should they have bought EqualLogic instead of Intransa, iStor, LeftHand or SANRAD?  The answer is no - just because Dell bought EqualLogic doesn't invalidate your choice.  You made that choice for a reason.  If you are worried that these other vendors will go out of business - it won't be because of this event.  As I said, at first there will be a rising tide.  If anything - unless the other iSCSI vendors screw it up - this could be a really good thing for them.

- There is a real opportunity for an emerging iSCSI vendor to become the thought leader in this space.  One could argue that there is plenty of room for a smart, aggressive, nimble smaller guy with a great solution to become an iSCSI category leader.  If their business is sound - they should race to go IPO and use the ~ $100 M they raise to knock walls down and go on a land grab. 

- Another giant may acquire one of the emerging iSCSI vendors.  The LeftHand customer asked me if they were likely to get bought out by "HP or VMware".  That was a compelling question.  And interestingly I was thinking along the same lines.  HP probably wouldn't buy LeftHand because their existing storage businesses would see it as an internal competitive threat.  Even if HP did buy LeftHand there would likely be so many political battles that it would diminish the chances of it being successful. 

VMware would be a very interesting acquirer.  This move would send shock waves throughout the industry.  Every other storage system vendor would be upset and concerned.  But I think it would be smart of them on a number of levels to make such a move.  Besides VMware's entire value proposition has been to change the status quo.  Hey, they took on Microsoft and are not only surviving but thriving. 

EMC might want to take a look around as well.  They need to be honest with themselves and ask - "is what we have in our product portfolio going to beat Dell / EqualLogic or do we need to get past the Not Invented Here (NIH) syndrome".  It is ridiculous for them to have an NIH attitude since CLARiiON was not invented by EMC and is now their most successful storage product.  In fact, if they didn't acquire Data General when they did and kept all of their eggs in the Symmetrix basket - as some of them wanted to do - their storage business would be in big trouble right now.  Perhaps they are at a similar crossroad.  EMC is the storage leader but they have failed to become the "Cisco" or "Microsoft" of storage with no where near 60-70% market share.  This was once a real opportunity for them but right now this seems like it will never occur.  Certainly, not with what they have today - because if it was going to happen it would have by now. 

HDS, IBM, NetApp and Sun should also be introspective and consider that maybe these next generation storage systems really have something valuable that they do not.  The key is whether they can integrate any new storage system into their organizations heathily. 

The final analysis - this is a good thing for iSCSI and end users.  FC is not going away any time soon - but you have to agree that Dell acquiring EqualLogic is a major milestone that raises iSCSI to another level.  It is a good thing for end users because this further validates iSCSI as a viable SAN technology and helps to mitigate the "risk" of implementing an emerging technology.  I think we can safely say that iSCSI has graduated beyond "emerging" based on a number of events, progress, advancements and overall success. 

I love the fact that Dell bought EqualLogic. It is bold. It is smart. And it makes a strong statement. Here is my quick analysis:

1. This enables Dell to potentially create and dominate the SMB storage market with a proven product. The SMB storage market is practically non-existant. What it needs is the combination of the "right" product and a vendor with enough brand awareness and resources to make storage networking pervasive in the SMB. This is easier said than done - but Dell and EqualLogic creates a compelling combination that has the real potential to make this happen.

2. Dell has told the world they want to own intellectual property - this puts a very big stake in the ground towards that objective. EqualLogic is not a "me-too" product - there is real differentiation. Additionally, it is field-proven with thousands of implemenations being used in production environments. Therefore there is no real risk in the product or technology.

3. Dell has also stated that they want to build a channel business - since EqualLogic is channel-based - and have actually built an excellent channel program - this helps Dell towards their channel plans with a differentiated solution.

4. VMware is changing the server business by driving consolidation of physical servers, which will have an inevitable negative impact on server revenue (no matter what anyone tells you). However, server virtualization actually increases storage networking implementation, adoption and revenue. Not only that but the margins are much better for storage systems than servers. EqualLogic gets 65% gross margins - that's really good. Additionally, storage is much stickier, if you will, than server technology.

The challenge for Dell is to turn its ~ $100 million EqualLogic revenue into a billion dollar business This will require considered and reasoned planning, attitude and strategies. Some of the people in the channel may reject working with Dell because they have been competiting with them for so long. Dell will need to navigate this with care. I believe that the strongest EqualLogic channel partners will stay that way - if they are assured that they will continue to be supported. If they actually thrive under Dell - it could enable Dell to create credibility with the channel.

EMC will be unhappy and regardless of what Dell or EMC say publicly - this will effect their relationship. Will it impact EMC's business with Dell- probably not in the short term. Even though EMC does support iSCSI, they have a strong FC-based business and this will continue to thrive within the world of Dell. Also - Dell has built a very healthy storage business with EMC and neither of them want to slow down this momemtum - they will keep selling CLARiiONs and Symms together treating it with business as usual (for now). Additionally, Dell will probably focus its attention on accretive storage markets - the SMB arena - going head-to-head with HP and IBM - their real competitors.

However, EqualLogic is not just for the SMB and certainly is not a low-end storage system. In fact, the EqualLogic (now Dell) storage system can scale to a massive solution with extremely impressive performance numbers. Additionally, it has sophisticated features and capabilities providing high availability, utilization, reliability and recoverability. However, these attributes don't make it complicated. Rather, one of the greatest competitive differentiators is that it is easy to manage regardless of how big it gets or how you use it. Therefore this acquisition has a lot of potential on many levels.

There will be a number of skeptics that will bet on Dell screwing this up. The naysayers should consider the following - this is a really big deal for Dell. Not just because storage is important to their business. Not just because they want to build a channel business. Not just because they want to prove to the world they can become a company that effectively sells technology that they own. Not just because they need to have higher margin businesses and differentiated solutions. It is because this move is part of a new Dell. I believe we are seeing the first phase of the next generation Dell and this is just the beginning. That is what this is about and that is what is at stake. So who wants to take bets that this will succeed or not?

I received a number of comments on my Infrastructure 2.0 blog. Here is one from Jay Krone, he is an EMCer and has some interesting comments that I would like to discuss:

Tony,

Your September 18 blog post, Infrastructure 2.0, posits that the future quite probably looks like reaggregated storage and servers, enabled by server virtualization software such as VMware. While I agree that such a move will ?completely change the status quo?, there are forces to consider that will affect the pace of this change.

1. The mainframe. As with many developments in IT, where the mainframe has supported it for 20 years or so, mainframe users have basically had CPU virtualization for decades, and the storage sharing that comes with it. The ?V? word has been part of IBM?s operating systems lexicon since just after OS/390, and IBM supports tools for managing resources in that environment. After all, isn?t an LPAR just a virtual machine? So Infrastructure 2.0 is probably Infrastructure 0.2, shipping today on the z series.

2. The hardware. The vast majority, probably over 90%, of the storage systems shipping today, shared or not, are purpose-built hardware. Some high end and midrange systems incorporate custom silicon, usually to improve performance per dollar. Systems built from ?commodity? components feature unique designs that optimize availability, reduce cost or improve performance. A design that makes a good server (memory and MIPS) and a design that makes a good storage device (I/O) are not the same and don?t fit economically in the same sheet metal. So until the price of semiconductor chips drops to just above beach sand, purpose built hardware will have a role.

3. The software. This, of course, is where the real challenge lies. Similar to storage hardware, most storage operating environments are purpose-built because they perform near real-time functions. Today?s virtualization environments generally are set up for data processing, not for I/O. And, unlike the mainframe environment discussed above, today?s up-and-coming virtualization wave is built on technology from many places, such that there isn?t a complete set of tools to build the Infrastructure 2.0 environment. For example, most virtualization environments have no software analog to the sharing performed by a storage network today ? that?s why they work so well with network storage. As for scaleable file systems and data clustering technology again we see nascent developments that hold a lot of promise at a higher level of abstraction than today?s storage products. But, for example, these techniques use brute force mirroring instead of the more economical RAID used by today?s storage products to protect data.

So, the Infrastructure 2.0 concept is very appealing. But as long as customers continue to value price, performance, and especially their data ? i.e. high availability, data integrity, and disaster recovery features ? it will be quite a while before we see all the pieces of Infrastructure 2.0 assembled into one enclosure. Especially in today?s open systems/open source world that doesn?t have the design by fiat luxury of z/OS. To mix two aphorisms ? it?s a small matter of herding cats.

Jay Krone

Here is my response:

Jay,

You mention that these challenges will impact the pace of this change. I agree and disagree. On some level it is happening already but I do agree that in order for it to become pervasive will require further innovation and development. Here are my thoughts in answer to your points:

1. I am really glad you brought up the mainframe comparison. What VMware has effectively done is leaps and bounds more powerful than the mainframe in terms of market impact. However, using the model of the mainframe - probably one of the most effective and efficient platforms in the data center - we can begin to get to mainframe levels of excellence within open systems. Virtual machines and at this stage -really VMware - has all the advantages of being on open systems (thus removing the fatal flaw of the mainframe of being proprietary) with the potential of having the same level of control regardless of the O/S platform. You get commonality of function across the entire data center (theoretically). This is what raises it beyond the mainframe - or at least it has the potential to do so. We always talk about how great mainframes are but in spite of this they are not and never will be dominant in the market. You see, that is the big difference - you potentially get all of the value with virtual machines and Infrastructure 2.0 without the big no-no of mainframes - being proprietary.

2. If you need specialized hardware then by all means use it. You said it yourself - "Systems built from ?commodity? components feature unique designs that optimize availability, reduce cost or improve performance". No one said that you couldn't have specialized hardware supporting storage systems that run on open systems platforms as a part of Infrastructure 2.0. We are not disagreeing. However, I do contend that there are plenty of storage systems that don't have custom hardware and the market share for these systems will continue to grow. Also - I am careful to say "specialized" - you can use off-the-shelf components to enhance storage architectures.

It also depends on what you are trying to get done. You are thinking of one size fits all storage. There are some storage environments that run purely on open systems server platforms with no specialized hardware. And they are doing a number of applications - including mission-critical stuff. Again, that doesn't mean that specialized hardware can't be used to optimize the environment - of course it can. But it certainly doesn't always have to - there are no absolutes. This is happening. Today. With thousands of users implementing the stuff. And EMC will inevitably follow suit sooner or later. I would be willing to bet you money on it.

And let's be creative here. Who says that the server has to become the storage system? Maybe the storage system will become the server. Why not run different business applications on the storage systems? Crazy as that sounds it is beginning to happen. It might make more sense for databases, for example, to run on the storage system with the user application still running on the servers.

3. LeftHand is now running their software in virtual machines and there are a number of other storage system solutions that are built on a Linux kernel or BSD. They could do the same. Additionally, now that VMware and Xen can run the hypervisor within processors, this opens up new possibilities since performance will be much better. Again, we can still use hardware assist if needed.

We often approach a problem with "no" listing all the reasons we can't do something and transcend the status quo. Personally, I don't think focusing on why we can't do something is the right posture versus figuring out what we need to do so we can make it happen - in spite of the challenges. By saying "yes" and overcoming the obstacles is how we get to new heights.

Tony A

If you haven't figured it out ? virtual machines have a direct and powerful impact on storage networking. We recently did research specifically on this subject. We completed our research this last August, which included surveying 706 companies and organization across different industries, countries and segment sizes.

Some interesting information came out of our research on how virtual servers will impact storage networking including:

• The amount of virtual server storage capacity on the early adopters' storage networks will increase by 23% over the next 18 months. What this means is that people are reducing the amount of DAS that they have and are putting more of their data onto network storage as a direct result of implementing virtual servers.
• The reasons that users increase storage networking based on their virtual server environments is the enablement of mobility of virtual machines between physical servers (66%); easier and more cost effective disaster recovery (61%); increased uptime and availability (56%); it's easier to upgrade physical servers (55%); and the ability to store multiple copies of virtual machines images for high availability (54%).
• We found that more than 70% of adopters will buy new storage systems. As adopters expand their virtual machine implementations they also grow their storage networks accordingly (63%); they need larger storage systems to accommodate consolidated server data (56%); and they want to use storage systems that are easier to manage for their virtual machine environment (48%).
• The top challenges that adopters had specific to virtual servers and storage included performance (51%); general lack of information and best practices (35%); and the need to perform more testing and qualification (31%) with storage.
• We found that majority of end users plan to implement new storage systems versus leveraging existing solutions. Additionally, many respondents also saw this as opportunity to look at new solutions from different vendors beyond their incumbent.
• Another interesting and important data point we uncovered was that nearly half of server virtualization users actually implemented remote replication as a part of this initiative. In other words, the adoption of server virtualization was a key driver to implementing disaster recovery utilizing remote replication technology.

I've rarely seen anything like the adoption of virtual server technology. Virtual server (and desktop) technology provides a no-brainer value in terms of server consolidation, reduced foot print as well as power and cooling consumption, and makes managing server environments much more fluid. From a visionary perspective, it is a leap toward Infrastructure 2.0 (see my last blog) and a highly virtualized data center from end-to-end. This is a major shift that will have far reaching implications ? perhaps more than anything in recent high-tech history (which is similar to dog years).

In this corner are your servers running your user and business applications. In the other corner are your storage systems. In between these two corners is an expensive and complex network.

Once upon a time there was no such thing as a storage network. Server and storage were bolted together. But you couldn't share resources. And since storage was so expensive, it was essential to share it. Additionally, you couldn't get a lot of capacity within a given footprint. So we decided to disaggregate servers from storage and put a network in between so we could share this precious resource.

By solving one problem we created a bunch of others. As a result, we live with lots of complexity, tuning, managing proprietary devices and networks, moving data efficiently, ensuring protection levels, etc. And of course there is a lot of cost associated with all of this infrastructure.

Perhaps it is time to consider gluing all of this stuff back together again. Why not run user applications and data management within the same infrastructure? You can have dozens and even hundreds of CPUs in a single rack. You can support 100s of TBs and potentially a PB of capacity in a single rack. Servers are made up of CPU, memory and disk capacity. Storage is made up of CPU, memory and disk capacity. Why aren't we consolidating these resources?

The problems that disaggregating infrastructure once solved are beginning to go away. If you put 20 virtual machines on a single server blade and all of it can access the same storage ? you don't have the sharing problem any more. You can get a lot of CPU and storage in the same physical foot print ? so that is no longer a problem. The cost of this infrastructure is magnitudes less expensive compared to when we decided to disaggregate. Additionally, you can use extensible file systems and storage clustering technology to share any and all capacity even if it isn't physically located in the same rack space. Consider combined Server/Storage systems that have as much (or little) CPU and Capacity as you need when you need it. A shared set of physical infrastructure that user applications and storage applications float on top of. You need more CPU ? add it to the pool. If you need more capacity ? add it to the pool. If you run out of space in a single rack ? no worries ? this is all highly virtualized ? grab more of these resources from any rack in the data center and add it to the pool.

The ingredients all exist to fundamentally change the way we implement our IT infrastructure. At the heart of this is the rapid and pervasive adoption of virtual machine technology. Add to that bladed architectures; dense disk drives; data clustering technologies; scalable file systems and faster CPU and you get Infrastructure 2.0.

Early adopters are already doing this today. Some vendors are beginning to realize the possibilities. This is a major shift and we are barely on the threshold of this new landscape. But it is looming. Infrastructure 2.0 will completely change the status quo.

It's a virtual world that is growing

Because the value is clear

VMware is on top

Their competition is nowhere near

VMware is ablaze

We are truly amazed

It's a big virtual world after all

* * * * * * *

VMware is driving storage networking

That is why I care

Our research shows that

Its driving up market share

The data center landscape is changing fast

If you're slow you'll end up last

It's a big virtual world after all

* * * * * * *

If you can't see the signs

This is the start of many things

VMware has got the world in its palm

And they soon will be king

The game has just changed

You've already been re-arranged

It's a VMware world after all

Merriam-Webster dictionary added some new words in 2007, one of which is ginormous.  The definition of ginormous is humongous.  Please note that the word humongous was adopted in the late 70s as a real word.  The word "ain't" wasn't a word for the longest time.  I remember being told by countless teachers that "ain't" wasn't a word and we shouldn't use it.  Well, that ain't true anymore because ain't is a word.  Another 2007 new word is "smackdown".  I can put these new words to use immediately: The battle for new ginormous storage systems will cause a smackdown from one vendor over the others and it ain't going to be pretty. 

There are some new words that are pretty cool like "hardscape", which means structures that are found in a landscape.  I can use this in a storage context as well:  The hardscape elements within the data center including storage, servers, switches and appliances are blurring in terms of componentry. 

At ESG we invent and promote new words within the IT world all the time.  The word hardscape conjures up an obvious one - softscape.  Witness the use:  The symbiotic relationship between the hard and softscape elements in the data center still holds true but its fundamental nature is changing. 

It is human nature to change.  That is why we add new words.  That is why we invent and create new technologies and products.  Some of this change may be gratuitous.  Others are driven by need. 

There actually will be a number of ginormous storage systems (let's have fun and call it g-storage) coming your way within the next 12 months.  These storage systems will range from 100 TB to PBs within a single system.  They will redefine the cost game built for very deep capacity. 

What will you use these g-storage systems for?  Answer: Whatever you want.  Mostly g-storage will be used for secondary storage as part of a Tier2Tier (T2T) storage (I just made that one up too) environment - nearline, digital archives, D2D backup, etc.  Why does the market need this?  Because we are seeing companies and organizations trending towards storing more and more data and they are keep it essentially forever.  Ginormous storage is an important next wave that enables new ways to use our information because we can store so much of it one place; and we can "act" on that information. 

It is important to note that g-storage fits within a T2T storage construct but not just as cheap storage (although for some that will be enough) but as an Enhanced Tier of storage. 

Enhanced g-storage will also be used for primary applications as well and not just as a secondary tier.  These applications include almost anything with the exception of traditional backoffice - highly transactional stuff.  Obviously, web-based applications, file sharing, something I call content mining, are all well suited for Enhanced g-storage. 

The requirements for Enhanced g-storage will include a number of elements.  Here are some things I think should be a part of an Enhanced g-storage system:  Ease of use, scalability and standards-based protocol support.  Other important considerations include capacity optimization technologies (e.g. snapshots, data de-duplication, thin provisioning), spinning down drives (e.g. MAID); and remote replication.  Enhanced g-storage is file and/or content aware.  Integrated search and indexing is also extremely useful part of Enhanced g-storage since these systems will have ginormous amounts of data.  True n-way clustering is an important element for ease of use, scalability, availability and linear performance growth.

All of the capabilities that I described above provides intelligence and enhancements offering greater value both from a device management and utilization perspective AND getting the most out of the information stored on it. 

Enhanced g-storage is coming and may one day be a part of your data center hardscape.  Enhanced g-storage fits into a bigger vision that ESG refers to as Infrastructure 2.0.  I'll talk about that in my next blog. 

The word out in the street is that storage spending is slowing.  But is it?  Some of the large storage vendors may be saying that and they may be right.  There is another thing to consider. 

There are over 50 storage system vendors in the market.  And some of them are doing really well.  If you add up the revenue that these upstarts are generating collectively you can easily exceed $1 Billion.  That is $1 Billion that would have normally gone to the leading storage vendors.  And that number will likely just get bigger and bigger.

I am often asked about consolidation in the storage system space.  Why hasn't it happened?  When will it happen?  Certainly there is some level of consolidation but it hasn't happened en masse.  I am not convinced that it ever will.  That is counter intuitive to the one of the laws of high tech - the rule of three. We used to believe that there was room for only three leading players in a high tech segment once it reached a certain level of maturity.  However, the world has changed.  The Internet provides us with more exposure and access; the emerging vendors are smarter; the end user customers are more savvy and don't just buy from the big guys.  Additionally, I think that storage is so diverse that there isn't just apple-to-apple comparisons and me too products. 

It doesn't hurt that the public markets have opened up again giving the storage upstarts a path to raise more money and awareness.  I was asked the other day by a reporter - why do these guys even want to go public?  It gives them greater credibility and perceived stability; it is a great PR event that you can keep riding; it can be a real morale builder for their employees driving momentum; and they can raise a ton of money to help get them to the next level. 

Over time some of the upstarts will go away either through acquisition or dying on the vine.  Some have and will go public and continue their journey of climbing Mt Everest.  But make no mistake, the old rules don't apply in the storage system arena.  Dan Warmenhoven, CEO of NetApp told me that he believes that the storage industry is 10 years behind the network industry in terms of maturity.  He predicts that we will be down to 2 key storage players within the next 10 years.  He may be right.  Right now I don't see it. 

Regardless, the upstarts are having an impact.  This year it will be over a billion dollars.  Next year?  You can be assured that the impact will be even bigger. 

What happened to CAS? For a while it was one of the hottest topics in the boring storage world. We could finally talk about content and compliance and not just speeds and feeds.There was a sense of urgency around CAS because of its relationship with compliance.

Coming up short when regulators request files and records can have major ramifications including hefty fines or worse. CAS solutions provide Write Once Read Many (WORM) technology and, as such, can offer immutable proof that a company's data has not been altered. Compliance is still an issue but it is no longer driving the same accelerated growth of CAS it once did. From a relative perspective the CAS market isn't huge, with less than $1.5 Billion in overall annual revenue. However, other far-reaching and sustainable market dynamics above and beyond compliance come into play in the CAS arena. The simple fact is that most companies and organizations create more unstructured data, including files, presentations, spreadsheets, images, graphics, etc., than any other data type. Additionally, it is becoming commonplace to create audio and video files, which consume massive amounts of capacity, even in mainstream companies. The majority of stored data is unstructured and the ongoing creation, storage, access and use of this data will drive the CAS market going forward.

Turning Data into Information

Consider the definition of data, in this context, as blocks that are stored onto disks. Today, this is a huge market, with hundreds if not thousands of applications reading and writing data onto blocked-based storage systems. The myriad of applications are each individual containers with little or no commonality between them. Further, the information that these applications create is essentially invisible or rather is unknown to the storage system or disks on which it resides. To the storage system, information is an electronic abstraction of that information, which we call data.

One of the most valuable ways that CAS transforms data into information is through federated search and indexing. Some CAS and NAS storage systems can find the content and information you're looking for by searching keywords based on filename and other metadata attributes but can also scan within documents for keywords. In this way, the quality of your searches is vastly improved and you are far more likely to find what you are actually looking for.

The name of the game is accessing the information you need quickly and precisely. Otherwise, you can easily be drowned by a sea of content. Or, conversely, you may not find the essential information that you need. It is important to have all (or as much of) your information in one place. This ensures that you are looking in the right place for what you need. It is non-productive to spend time looking for something when you are not sure whether it is there or not. Federated search enables leaps in efficiency since it creates commonality where none existed. It enables you to leverage all of your content independent of the various applications.

Some NAS and CAS solutions provide federated search and indexing using metadata and within content, raising it from data to information storage. In the case of compliance, search and indexing can save you from a hefty fine. It could provide evidence to support your case during litigation proceedings. During a project, intelligent search could save you hours, days and perhaps weeks. These are reactive events that are potentially critical to your business. Additionally, a storage system with quick and precise search capability can be used proactively to help you to research new product or service ideas and create businesses. Can your current SAN-based storage system provide this type of functionality?

CAS, NAS and SAN and the NNG

The IT world is moving from the structured data (database) archetype to an environment dominated by unstructured data (files, images, multi-media). As this shift occurs then new priorities and requisites will transform storage systems from data to information solutions. The next, next generation (NNG) storage system must be great at everything and at the same time easy to use. The NNG storage system must move data quickly (e.g. SAN) and understand content (e.g. file, NAS, CAS) and provide applications that turn data into information (e.g., search). CAS will no longer be a separate category but a part of the overall system. This will be true of SAN and NAS as well but not as different protocols or controllers that work independently of one another but an amalgamation of all these technologies.

Marc Farley, the author of Storage@Work - a EqualLogic blog - just wrote an overview on thin provisioning (The Basics of Thin Provisioning).  In the blog he quotes Eric Schott, director of product management for EqualLogic, comparing thin provisioning to credit debt:

"...newcomers to thin provisioning should think about it as being analogous to credit and should try to understand what the potential down sides are before starting to use it. To be specific, they should understand that at some undetermined point in the future they will need to make good on the 'capacity debt'."

I like this analogy except that it is somewhat inaccurate.  For one, thin provisioning is a debt that you may never have to pay back.  For example, if you buy a storage system with 10 TB of capacity and never consume all of that capacity then you never have to pay back the capacity "debt".

Taking the credit analogy further = - there is no "interest" payment with thin provisioning.  In fact, you can save money with thin provisioning since you can delay the acquisition of capacity - you can buy to your needs and not to your forecast.  Since the price of capacity goes down over time - you end up paying less for the same amount of capacity buying it later.  This sounds more like trading commodities - an investment and not debt. 

If you consider thin provisioning a debt - it is an interest free debt that can actually save you money and in some cases you may never have to pay it back.  Who wouldn't be interested in that? 

The Skinny on Thin :-))

As predicted thin provisioning is being more widely aodpted by a number of storage vendors of all sizes.  Let's layout the landscape:

3Par: They are very focused on thin provisioning and in many ways they are the vendor most closely associated with this technology.  They've had it for years with lots of customers using it in mission-critical environments and arguably the technology and revenue leaders of thin pro.  They've also done a great job creating awareness around thin pro - maybe a little too much.  3Par has a great storage system - in part because of thin pro - but thats not the whole story. 

Compellent:  Like 3Par, these guys implemented thin provisioning as part of the core feature set and it is fundamental to their design.  Like so many of the guys that support thin pro - they just don't make enough noise about supporting it.

EMC:Yes, they do support thin provisioning in their Celerra NAS solution.  The challenge of being a big guy is that you have so many products that often a single feature gets lost in the shuffle.  And Celerra is not as widely deployed as EMC's SAN storage products - so the market impact is not as big. 

LeftHand:  Did you know that LeftHand has more implementations of thin provisioning than anyone?  3Par has generated more thin provisioning revenue and awareness but in terms of sheer deployments LeftHand is the leader.  They have over 2,000 end user companies using thin pro. 

NetApp: Thin provisioning is a part of FlexVol.  NetApp was the first leading storage vendor to support thin provisioning but I don't think they've leveraged that fact in their favor as much as they could.  Of course, end users are buying NetApp for a lot of other reasons but associating themselves with thin pro is additive value. 

Do you see a reoccuring theme?  Only 3Par is adequately educating the market on thin provisioning out of this bunch.  And if I didn't mention you - don't get mad at me - get mad at yourself. 

New Entrants:

EqualLogic: They have jumped into the thin game.  Thin provisioning fits well into their vision/strategy.  Since they provide thin provisioning as a part of the system (like all their features) at no additional cost they should have a bunch of depolyments in production pretty quickly. 

HDS: They are making a big deal about thin provisioning in their USP V.   They are one of the leaders in SAN storage and in many ways they validate this technology - especially since they are being so vocal.  This will help all the other guys that support thin provisioning and it should drive their competitors to support it as well - which brings more value to the end users. 

The HP XP24000 is the OEM version of the HDS USP V, so they support thin provisioning at the high end as well. 

Sun:  They resell the HDS USP V as the Sun StorageTek 9990.  I assume that they support thin provisioning at the high end but they've made no noise around it at all.  You can't find any information on their web site or in the Google-sphere. 

Thin provisioning is at a point of acceleration.  More vendors are supporting it; as a result more end users will become aware of it and deploy it; other vendors will begin to support it to address competitive pressure; and then even more end users will adopt it - rinse and repeat.  Over time thin provisioning will become pervasive.  We are at the threshold of this happening.

My other blogs on thin provisioning:

Thin is Phat

Thin Provisioning: Confusion Abounds

We are about to embark on new research focusing on virtual server technology and its impact on storage networking.  I believe that the impact will be profound. 

A great deal of storage is not networked even though the fundamental reasons including data growth, consolidation, centralized data management and protection have been there for years and years.  Virtual servers is a viral phenomenon that will be a catalyst for essentially universal storage networking adoption.  This will cross industries and market size.  Why?  Because companies and organizations have embraced the value of implementing virtual servers to better manage their server infrastructure.  True consolidation is happening at the server layer.  Once you have multiple virtual servers consolidated into a single physical server it becomes impractical to have internal storage.  Beyond the impracticality, it also becomes imprudent.  Having multiple virtual servers on a single physical platform creates risk, which can be completely mitigated by storing virtual server images and their corresponding applications and data on a reliable storage networking platform.   Additionally, maintenance tasks are greatly simplified such as physical server upgrades, bringing new applications online, and the ability to transparently move virtual servers to other platforms based on performance needs.  These day-to-day tasks will be greatly simplified and provide a leap in efficiency (and we like leaps).  Other value is created including reduced floor space; reduction in power and cooling; and the ability to implement cost effective disaster recovery.  The benefits are far reaching. 

The above realities not only deal with large or medium-sized shops but across the board even in small companies.  In fact, they may find even more benefits because of their limited resources.  They can set up a highly available environment leveraging virtual server technology and networked storage.  Additionally, they can take advantage of the management value of this type of set up, which is a major need based on resources. 

I've had a bunch of conversations with end users of all sizes on this that confirm my thesis.  They consistently tell me that they are going to network all of their storage based on their virtual server implementations; and that they are going to leverage virtual servers as a part of a disaster recovery strategy.  We are on the threshold of something very big. ESG is going to set out to quantify it in detail.  Stay tuned. 

I just read Marc Farley's blog on FCoE (FCoE: Run Away, Its the Monster) and felt compelled to write about this subject as a result.  I really like Marc's passion but I can't say that I agree that FCoE is a monster.  My initial reaction is that FCoE is a good thing.

FCoE - or FC over Ethernet allows you to use Ethernet to build FC SANs instead of - or in conjunction with FC infrastructure.  The value of FCoE is that it perserves WWN addressing, zoning parameters, multipathing, SMI-S, etc already established with existing FC SANs.  It actually makes it easier for FC-centric end user shops to leverage Ethernet.  You will be able to use FC switches by adding FCoE ports.  It will be Ethernet "in" from the host systems and then FC "out" to the existing storage systems that support FC ports.  It allows the FC storage adminstrators to still be in control of their own fate versus having to go to the networking guys for infrastructure. 

Additionally you be able to have FC and iSCSI traffic (as well as all of your other network traffic) sharing the same Ethernet infrastructure.  That is a good thing.  FC and iSCSI will be able to share the same front end of the network happily moving across common Ethernet networks.  In fact, FCoE may actually help iSCSI gain a greater integration over time.  FCoE can be a transition point from FC to iSCSI. 

One of the big decisions that storage system vendors will have to make is whether they will support native FCoE ports in the future.  If yes, then FCoE has a chance of being more viable over the long term.  However, if the storage system vendors feel that the extra development and Herculean interoperability testing effort isn't worth it, then iSCSI will ultimately replace FC at the high end - just as it is doing in the mid-tier of the market. 

It is important to remember two things.  First, any new standard and technology takes much longer than anyone supposes to have a real market impact.  That will be true of FCoE.  Second, iSCSI provides greater benefits than just running over Ethernet.  iSCSI and the IP protocol are much more virtualized than FC - making it much easier to manage and scale.  FCoE does provide value by leveraging the existing FC ecosystem but it has nearly all of the limitations of FC:  You still need specialized equipment; FC expertise; it is more distance limited than iSCSI, and it requires more manual management. 

Like so many things FCoE is not a panacea but it could be a good transitional technology that helps FC-centric data centers save on some costs by sharing Ethernet infrastructure while perserving their FC investments.  There are four possible outcomes for FCoE (that at least I can think of right now):

1.  FCoE could be a non-event and gain no traction.

2. It could help pave the way for iSCSI to be dominate at the high end - FCoE would be transitional as a step towards iSCSI.  It perserves the FC ecosystem but gets the data center sharing much of the same Ethernet infrastructure.

3. FCoE could replace FC at the high end if storage vendors embrace it and if there is still a bias against iSCSI at the high-end.  This will be timing more than anything.  As iSCSI becomes more mature, there will be less of an issue with adopting it at all levels.  For FCoE to gain any real traction, the storage system vendors will need to support it quickly. 

4. It will be the dominant SAN interconnect of the future.  I wouldn't bet one dollar on this one coming true. 

FCoE does support something that Ethernet inventor Bob Metcalfe said to another ESG analyst, Brian Garrett - "I don't know what comes after Ethernet, but it will be called Ethernet". 

I was actually ready to talk more about FAN (File Area Network) and how Cisco has joined the fray and was about to go toe-to-toe with rival Brocade on this emerging segment.  And then Cisco decided to stir the waters a bit (if you don't know what I am talking about, google it).

First, it is a brilliant marketing strategy because other than the initial acquisition, NeoPath would otherwise have received no further ink.  But we all like a little controversy.  It isn't quite on the level of Brittany Spears doing something any young, ungodly famous, drunken, unbelievably rich and idiotic person would do in public - but in our own little storage microcosm we take what we can get.  However, I doubt that this was intentional on Cisco's part. 

What is really of concern is the whole segment of NAS or File Level Virtualization.  It raises an eyebrow that these companies - NuView, Rainfinity and now NeoPath have been acquired purportedly for such little money.  This is an important indicator of the overall value of the market.  The board and the management teams of these startups came to the conclusion that it was better to be acquired for relatively small sums versus trying to build the company and market.   But then Brocade and EMC, two of the acquirers, will tell you how strategic these solutions are to them.  And they may very well be important to them but neither of these major vendors were willing to pay a lot of money for these "strategic" solutions.  Good negotiations or good BS?  Tbd.

But maybe there is another problem.  How well do these solutions really work?  They may be equal to the claims of the vendors but then again, they may not be.  Maybe this stuff is inherently complicated.  Or perhaps some or all of these products still need a lot of improvement.  The devil is in the details.  The movements of the market indicate that there is something amiss.  We must also consider that NetApp, the king of NAS, has for all intents and purposes ignored this whole NAS/File Level Virtualization thing.  They are smart guys and know this business better than anyone.  NetApp has been pretty ho-hum about this all along. 

Microsoft would be wise to step into this battleground.  They do have DFS but that is really more of a technology than it is a product.  Brocade StorageX (NuView) and EMC Rainfinity use DFS leveraging the global name space.  These solutions provide easy to use tools, policies, data movement and reporting around DFS.  If Microsoft provided a compelling File Level Virtualization solution to better manage its file server environments it could have a major potential impact on the NAS market.  And apparently, the NAS Virtualization vendors come pretty cheap based on the recent historical trend.  Bill Gates makes more money in two weeks than the cost of these acquistions.

There are two emerging vendors still standing, Acopia and Attune.  They are fighting the good fight but these meager acquisitions have to be a concern to them.  How do they elevate themselves beyond the mediocrity of their contemporaries?  No doubt they will pound their chests and say how great they are and how much the other guys suck.  However, this won't really help them.  Instead they should go down a different path (and definintely not the Neo "path").  Read on. 

Conceptually I am a big proponent of FAN (FAN is easier to write than NAS/File Level Virtualization - even though I just wrote it again- doh!)  I believe that there is no real visionary out there that understands what FAN can bring to the table.  The tactical objectives are to use FAN as a heterogeneous data mover creating an intelligent tiered storage environment; to provide a single view and access to all of your file systems and content to improve workflow; and simplify client side management.  You can do this for NAS storage and for file servers.  However, the main focus has been on the NAS side.  But there is huge potential for simplifying file server environments.  In fact, you could obviate the need for NAS if you simplify the management of file server environments by implementing FAN.

This last part is something that none of these vendors are willing to do.  They all want to play nice as an add on to existing NAS environments versus competing against the NAS vendors.  But the NAS market is finite especially when compared to the bigger storage network market.  Therefore as an add on to the NAS market - you are a smaller subset of a minority subset.  Take the tiger by the tail.  Be bold.  Certainly make NAS environments better but also (and perhaps more importantly) compete directly with NAS. FAN creates a virtual NAS solution out of your existing file servers and file systems.  This gives you three things to go after - as a NAS add on (which they focus on today); as a NAS competitor/replacement (which they don't do); and as a transparent way to manage existing file servers without implementing NAS - these would be SAN shops that have no NAS but a number of file servers (which they don't do and probably hadn't thought of). 

But then again - maybe FAN solutions aren't ready to take this challenge on yet. 

Even beyond competing with NAS - there is a bigger picture in the distant horizon.  The true convergence of storage protocols - an amalgamation of file and block combined with highly virtualized storage infrastructure.  That is a vision worth sinking your teeth into. 

Instead the FAN guys are underwhelming us with improving NAS environments.  Great.  That and $40 Million to $60 Million will get you acquired (and a cup of coffee). 

If you think about it NetApp is really the only major NAS vendor in the market.  EMC can make all the claims they want but they are not really growing their NAS footprint outside of core EMC SAN customers.  The other NAS upstarts - Agami, BlueArc, Exanet, Isilon, ONStor, Panasas, and Pillar are not competing against EMC.  But they are all competiing against NetApp. 

NetApp has owned the NAS market for 15 years.  They pulled a fast one on all of the other major storage vendors.  HP has no real NAS business.  IBM woke up and partnered with NetApp.  EMC at least has been able to keep NetApp out of a lot of its core customers.  Sun invented NFS but missed the boat.  Dell doesn't even really play.  HDS has just partnered with BlueArc and surprise, surprise they are selling NAS to their customers. 

Why doesn't Sun partner with a NAS vendor?  They bought Procomm in 2005 for their NAS solutions.  Some of you might be saying - Pro who?  Sun isn't really making any hay with their NAS (I just made a pun in case you missed it).  The SUN NAS solution doesn't have any significant capabilities above and beyond its competition (NetApp).  The only thing they can compete on is price and customer loyalty.  The former is not a desirable place to be and the latter hasn't been earned in regards to their NAS solutions. 

And what about HP with their recent acquisition of Polyserve?  Come here.  Closer.  Shhhh.  Polyserve isn't a NAS product.  Polyserve is a clustered file system that enables you to create server and file-based storage clusters.  How is that different from NAS?  There are table stake features that NAS products have such as file level logical snapshots.  HP won't win the battle by going head-to-head with NetApp.  That doesn't mean that HP can't win lots of deals with their Polyserve solution.  Just don't call it NAS.  HP should sell its leading EVA SAN storage system with file storage capabilities.  Its easy to use, cost effective and is extremely scalable.  They should go out and sell that to every EVA customer they can. 

The HDS / BlueArc move was a smart one.  HDS has large enterprise customers and BlueArc is well suited for that market.  Together they can build a strong high end business.  The next step is to go down market a bit.  But it is easier to go from high to low than the other way around. 

And what about Microsoft?  Absolutely kicking butt on Windows NAS but at the lower end of the market.  And they have tons of file servers out there. 

How about the upstarts?  They have a long way to go.  But there is opportunity out there.  And the upstarts all have compelling technology.  Some of them have great products.  And some of them also have great business execution.  They will need all three (and a little luck) to take it all the way. 

Right now NetApp is in the NAS driver seat (I could make a Nascar joke right here)  They are the kings of NAS regardless of whatever weird statistics are out there.  But even they are getting distracted.  They are doing so many things at once focusing on FC and iSCSI SAN; unified storage; VTL; the SMB market; the data center; on and on; that they are kind of forgetting about NAS.  Go to their homepage and you won't find one mention of NAS but you will find SAN.   They have a leadership position that they are not leveraging strongly enough to their advantage.  Bad timing too since file data is growing at such a rapid rate, has increased in importance, and consumes a ton of capacity. 

Here is a prediction that impacts every storage system vendor out there and every consumer of network storage.  At some point file-level awareness will be requisite.  That doesn't necessarily mean you need a NAS solution (but they are certainly file aware).  Why?  Because having file level awareness enables a ton of capabilities and end users can use the storage processors to execute those capabilities. 

This isn't just about NAS but about supporting higher levels of intelligence within the storage system.  It is about transforming data (blocks) into information (file/content).  It is about running applications that are resident within the storage system construct to use that information for the business versus just storing and protecting it.  It is about elevating the storage system to more than just a box with spinning disks, high availability and replication software. 

As a self proclaimed storage geek I have to confess that when I was younger I did collect comic books (and I know for a fact that many of you are in no position to throw stones).  In honor of that I invented two new super heroes - Invisible Storage and Strategic IT. 

Invisible Storage has the power to become invisible to the business.  And let's face it, the only time that storage is visible is when there is a problem with it.  If performance is too slow, data is unavailable or if systems are down, and when projects take weeks or months to come online then suddenly we are all too aware of storage.   Storage is also very visible at budget time.  You need to spend how much?  You need to have how many people to support this stuff?

We need to make our storage invisible to the users of it and ensure its perpetual optimization.  There are some end user companies that actually achieve this but at a high cost in time, resource and money.  We expect our smart people to take care of it.  But we can't just throw good people at the problem.  Why?  Because people don't scale.  And because humans are the inventors of human error.  No matter how smart your people are they will make mistakes.  We need to leverage technology to work for us and provide perpetual optimization.  And by its very nature, this process needs to be easy to be effective, thus removing the manual management that we lean on so heavily today. 

Strategic IT is pretty much an oxymoron with most end users.  Sure, IT is important but is usually considered over head.  But what if we lived in a world where IT was strategic to the business?  Not just to keep your email up and running but actually leveraged technology to make you more competitive and enable new services and products.  Instead IT are those techie folks that handle that storage stuff. 

IT is complicit in this perception of itself.  IT personnel like being techno geeks.  They like that they know the ins and outs of big boxes with lots of shiny lights.  It makes them feel special.  No one can touch them.  No one bothers them.  People walk by and say - "I don't know what he does, but don't bother him because he is the only one that knows what he's doing". 

That is where Invisible Storage and Strategic IT come into the picture.  One of the biggest problems we have in the data center is the complexity of individual storage systems.  Do you still manage individual LUNs?  Individual file systems? 

You need special virtualization powers - it's kind of like the force (ladies you can go home now).  When people ask me what is the big difference that the new generation of storage systems brings to the table that the last generation lack, it all comes down to how highly virtualized these systems are.  Virtualization manages physical assets - to deliver perpetual optimization. 

Storage level virtualization can be manifested in multiple ways.  Within the storage system it is the ability to treat various hardware resources as logical pools.  This can include CPU, memory, bandwidth and capacity.  Thin provisioning is a powerful form of virtualization that liberates system administrators from having to be bounded by the limitations of traditional provisioning.  Having large virtual volumes that can stripe data across dozens, hundreds and even thousands of disk drives within a single storage system for optimal performance - and without having to manage individual LUNs.  Read-only snapshots and writeable snapshots are one of the most useful forms of virtualization.  True clustered storage is another powerful form of virtualization that enables you to add n-way storage controllers that scale to massively parallel storage systems.  If you add all of these things up - you get the tools to implement a perpetually optimized storage system.

The point is that storage systems need to be highly virtualized solutions.  Those that are not highly virtualized will really be at risk as the new generation storage systems stake their claim.  These solutions are not just "me too" products but they are compelling solutions that solve the fundamental storage problem that still exists - cost and complexity.  Make things easier and less expensive makes storage invisible and frees up IT to be more strategic. 

Have no fear - Invisible Storage and Strategic IT are here to save the day.........!

Whenever I write about iSCSI I get a lot of feedback.  In my blog - The Year of the Boulder or iSCSI Hype and FUD - I was making a particular point to Chuck Hollis' blog - Where is iSCSI These Days, in which I argued that Chuck was being a bit negative about iSCSI.  ESG has done research and spoken to dozens of happy iSCSI customers and it is well on its way, momentum is unstoppable, etc, etc. 

I received a comment from someone that calls him- or herself the "anonymous coward".  I actually applaud the anonymous coward for his/her anonymity in this era of "YOU" narcissism (I know this sounds strange coming from an analyst that is in the public eye quite often).  But that aside - anonymous coward sent us this thought:

"IDC's 5% market share reported by Chuck isn't inconsistent w/ the 17% penetration reported by ESG's survey of 511 survey users.

Reasoning:  First, ignore the problem that the former data point is annual shipments and the latter is installed base.  With the iSCSI market growing so fast and being so young, this is an OK simplification.  Second, assume a significant percentage of users of iSCSI are still investing in FC.  Third -- and most importantly, assume iSCSI is a *lot* less expensive and requires a lot less SAN-specific gear because it relies in part on NICs and switches already deployed as part of the data network.

The implication is that the iSCSI equipment market is likely NEVER to be as large as FC at FC's peak.  If your business is to sell SAN gear, be afraid -- be very afraid."

First of all, my issue with Chuck's blog was with what I perceived was the negative tone towards iSCSI.  I didn't really care about the numbers.  But "anonymous" does make a good point.  Since iSCSI is less expensive the annual revenue will be lower than a comparable FC SAN would be.  However, I don't think that iSCSI is a *lot* less expensive.  The numbers would be somewhat less based on the iSCSI cost advantage but not staggering.  The biggest cost is the storage system.  If EMC or NetApp is selling the iSCSI storage system, it won't cost any less than their FC versions.

What I really want to talk about is the last paragraph in the comment.  Anonymous coward believes that iSCSI may never generate more revenue than FC at it's peak because of its inherent cost advantage.  A provocative analysis.  It all depends on what he/she means by "SAN gear".  If you mean the HBA's and switches - I agree.  But the cost of the storage systems should not change based on the interconnect.  And we need to focus on selling solutions and not gear.  The fact that we still measure pricing for intelligent, sophisticated and valuable storage systems at a price per GB is ridiculous.  And every vendor out there should realize that if you continue to add value than the money will come.  If an IT department has a million dollars in their budget - they will spend a million dollars.  If they have ten million dollars - they will spend ten million.  Just make sure they spend it on you by adding value and not overpriced nuts and bolts.  IT departments don't mind spending money but they do mind wasting it. 

I disagree that iSCSI won't surpass FC at its peak in terms of revenue.  FC has had great success but has failed to reach the level of implementation of LANs.  That is because the cost/complexity ratio of FC exceeds the amount of pain end users have with their current DAS environments.  iSCSI has the potential to succeed where FC has failed.  And this isn't just in small companies but companies of all sizes.  There are dozens, hundreds and even thousands of servers in large global organizations that aren't a part of the FC SAN.  That is why the concept of SAN Inclusion is so important - including stranded servers via iSCSI to the SAN.  There are also companies with between 1,000 and 5,000 employees that are still using DAS and have not implemented a SAN. There are hundreds of thousands of companies out there that have no networked storage.  By the way, all of these companies do have LANs and pretty much all of their servers are connected to it.   

The market for iSCSI includes replacing FC SANs over time as well as greenfield opportunities, both of which offer enormous growth opportunity.  That is why iSCSI should one day surpass FC in deployments and revenue.  Additionally, the market will change.  One of the most compelling aspects of iSCSI is that it flattens the IT organization.  Once all of our data is moving over the same network infrastructure - not the same network but the same type of network - new opportunities will arise.  For one, the convergence of NAS and SAN that combines the intelligence of NAS with the performance of SAN.  That is the next step in storage that has far reaching implications. 

What else will happen in the world of storage?  Here some more ideas continued from Part One:

6. SAN Inclusion - this is a concept we came up with that is related to iSCSI.  The general idea is to use iSCSI as a way to include departemental and workgroup servers into the SAN within the data center.  This was one of the big promises of iSCSI and we feel that now that the technology has entered the early mainstream.  There is a ton of value for storing all of your data on networked storage.  But we still have failed to do so to a large extent.  We intend to bang the drum more loudly on SAN Inclusion.

7. Persistent Data.  This is a term that a number of vendors have embraced.  I like it as well.  We focus on transactional data but up until now didn't have a strong term for data that doesn't change very often, if ever at all.  While Persistent Data has always existed there is now greater awareness and products targeted specifically for it.  Many companies still use a one size fits all approach to storage and keep all of their data on their most expensive systems.  This is crazy since anywhere between 65% to 85% of your data is never or rarely accessed again soon after its creation.  Move the persistent data off of primary storage onto a lower tier. 

8. More Fun with FAN.  This is file-based storage virtualization.  It makes a ton of sense - see my blog for more on FAN (Fun with FAN).  This technology should be integrated into any environment that has lots of files. 

9. Power and Cooling.  Expect to see EVERY single storage system vendor talking about how their technology helps you manage power and cooling.  Some of it is true and makes sense.  Others will be stretching things just a bit.  There are architectural design elements that can make a difference in power and cooling.  There are also some technologies that can make an impact such as capacity optimization (data de-duplication and other techniques), thin provisioning, writeable snapshots and spinning down drives. 

10. Clustered Storage.  This is a form of virtualization that the majority of people just don't get.  I think its a no-brainer.  Why cluster storage?  It provides just-in-time scalabilty.  Don't buy extra head room but grow as needed.  What?  You don't need scalability?  Clustered storage has another advantage over traditional dual node midrange storage systems.  If one of the controller nodes goes down you are at 50% performance.  With a three node storage cluster you would be at 33%; four nodes at 25%; five nodes at 20% and so on. 

I am anxious to see how the storage IPO market impacts the emerging vendors.  If companies like Riverbed and Isilon sustain their success this bodes well for other storage players.  Some will go public and others will get attractive buy outs.  The momentum will trickle down and the VCs will smell money and begin to invest again in startup storage companies and more innovation will be funded.  And maybe, just maybe being a storage wonk might become cool.  Maybe I could be interesting at parties.  Maybe I will get invited to parties.  I could have friends that like me and want to hear what I have to say.  I could get my own byline in the Wall Street Journal.  And...

I just read Chuck Hollis' blog on iSCSI (Where is iSCSI These Days?) and Dave Hitz's response to Chuck's blog (The "Year of iSCSI").  The EMCer (Chuck) and the NetApper (Dave) both had interesting things to say about iSCSI.  Since I have been known to talk about this subject I figured I would add my two cents. 

Chuck's thesis is that we have been talking about iSCSI for five years now and it should have entered the large Enterprise companies by now but it hasn't.  He implies that iSCSI should be further along than it is.  In both cases I disagree. 

Chuck should qualify his observations.  He only has insight to EMC Enterprise customers, which is not the universe.  There are Enterprise companies that do not have any EMC storage products.  We recently completed an iSCSI research project that surveyed 511 end users and we found that 19% of early iSCSI adopters are companies with 20,000 employees or more and 20% of planned adopters are in this same category. 

I for one hate the "Year of This" or the "Year of That".  It takes a long time for a technology to become dominant in a market and it isn't like flipping a switch - one second its off and then the next its on.  Success is not binary but rather comes in increments.  It took 44 years for the automobile to reach 25% adoption in the US.  That is because the market conditions needed to be right and an ecosystem needed to be built. 

iSCSI is on a multi-year journey.  It started with an idea; became a standard; technology was developed; products were made; the technology was further improved; ecosystem partners joined the party; end user education was required ("just what is this iSCSI stuff anyway?); early adopters needed to weigh risk against reward; more products came out; more ecosystem partners; more customers; more education and awareness; more customers; more products; more ecosystem partners; on and on and on.  Momentum is built if there is value and execution.  In the case of iSCSI, there is both. 

This process takes time and resource.  It is like pushing a boulder up a hill.  You can't instantly move the boulder up the hill and then claim it is the "Year of the Boulder".  You gain success with each passing day, week, month, year, etc. 

Chuck made some great points but in my opinion they get lost to some degree because of the somewhat cynical tone he sets the stage with.  His blog is less about iSCSI and more about his education on why iSCSI hasn't met his expectations.  It is a good education and is valuable to those who read it and don't get distracted by the negative undertones. 

Disparaging iSCSI is the other side of the same coin of over hyping it.  Both are as equally counter-productive.

Enough about Chuck.  Let's move on to Dave's blog.  I think he is spot on.  Except for one point.  He states that iSCSI is for low end environments.  Our research shows that this not the case.  iSCSI is being used for a wide range of environments spanning from a handful of TBs to hundreds.  It isn't iSCSI that is the issue.  It is the way that the storage systems and networks are architected that determine the scale. 

We need to educate with as little prejudice as possible.  That is the way we really improve things. 

Some other iSCSI-related Stor Wars blogs:

iSCSI - The "I's" Have It

iSCSI - Playing Nice in the SAN Box