The Year Winds Down
December 31st, 2008So much has changed, yet so much remains the same. I have been reading through the various trade pubs (on their way to the trash can) to see what others thought was significant about 2008 (and what wasn’t). I won’t go there in this column except to say the following.
The most important things, IMHO, that appeared in 2008 from a storage infrastructure standpoint were the following:
1. XIOTECH Intelligent Storage Element (ISE) — If these guys play it right (and I am doing my level best to help them do so), ISE is an extraordinarily important development that could change a lot about how we build storage today. ISE, you see, is a “no frills” Lego building block from which storage infrastructure can be built intelligently and in a manner that meets business needs. Hardware is hardware, but this little guy enables you to add the features and functions you want, instead of buying a flock of features you don’t use just because that is how the vendor is packaging them for sale. A lot of my clients are excited about ISE and even those who are locked into a three-letter vendor are looking to phase out the big iron arrays for these guys going forward — a phase-in strategy because rip-and-replace is untenable. It makes sense to me and provides an essential building block for creating a real SAN.
2. XIOTECH (again) Web Services based management. The implementation of a management platform based on Web Services standards from W3C is perhaps even more important than Casey and Steve’s ISE-box itself. Let’s do the math. Applications are already requesting services (that means hardware resources and software functions) from the infrastructure using Web Services requestors. Problem is, nobody is answering. While SNIA dicks around with SMI-S and XAM, the truth is that the real solution to common infrastructure management (and perhaps intelligent data management too) is staring them right in the face: Web Services. Through this standards-based approach, we could collect copious information about hardware and software services in a network — much more accurately and in a more timely fashion than SNMP traps can do it, and at least on par or more efficiently than accessing this info via a vendor API. And enabling gear and software to speak Web Services and support Java/XML formats for reporting is a heck of a lot easier than writing custom scripts to get at this info or using a difficult to code and implement SMI provider. That SNIA ignored an existing framework with so much potential — one that works already with Microsoft, Oracle, Linux and perhaps even the hypervisor crowd – confirms my view that they are yesterday’s news. Bottom line: I think Web Services-based management is a game changer. I strongly recommend that anyone who wants to plug their products into a common management scheme consider enabling their hardware/software for Web Services dialog. Xiotech can show you how.
3. Server virtualization has done us all a service. Yes, I know I have ranted here against the hype, but the truth is that the pain created by server virtualization (especially VMware) has forced many folks to take a systemic view of their infrastructure, starting at the application level. This is far superior to building storage infrastructure without careful consideration of what applications and their data actually need. The problems with management, with connections to so-called SANs, and the revelations of how much waste already exists in the storage we own are kicking a lot of stovepipe array makers in the buttocks. This probably wasn’t the goal of the virtual server software guys, but it has done us all a favor. We needed to get our minds right and the frantic embrace of virtual servers is teaching most of us some hard lessons about the crap that we allowed vendors to sell us over the past few years.
4. The more I learn about their next gen products, the more I believe that CA is finding its stride again. Mainframe 2.0 is a smart initiative. It’s not just about putting a user friendly face on the mainframe OS so that the Java/XML trained IT generation can get a clue, it is about making the mainframe a centerpiece of an efficient and economically managed infrastructure again. I also like what is happening with ARCserve, a product I abandoned some time back in favor of a competitor. I am back on it now. I like how CA has added functionality to the product, including a software VTL and software deduplication engine — not to mention some of the key functionality of their ill-timed BrightStor SRM suite. (I say ill-timed because no one wanted to buy SRM when money was plentiful and gear was cheap. Today, that has changed and ARCserve is worth a close look by shops large and small.)
5. I have warmed up to Crossroads Systems of late, particularly for their Read Verify Appliance and its reporting capabilities. This is a must-have for anyone who wants to make tape operations efficient again in their shop and to cope intelligently with the data burgeon that shows little sign of coming under intelligent management anytime soon.
6. I also like Virtual Instruments — a lot. From where I’m standing, these former Finisar-ites are doing what the mother company never could: they are telling the truth about the inefficiencies of FC SANs. To some extent it grates on me that they need to soft peddle some of the issues that they are uncovering with their management kit. It wouldn’t make them very popular with the big iron and switch makers if they normalized their data and told the truth about what an abyssmal failure FC fabrics have become, as evidenced by extremely poor allocation and utilization efficiency metrics collected FROM THE SOURCE. Instead, they have chosen (smartly) to take the high road and to message that they are merely a companion product for the big iron SAN that will help you get more value from your investment. True enough. But it stops short of telling the real story about port inefficiency, poor gear performance, the performance hit caused by a lot of the value add stuff that vendors are adding and we are buying even though it doesn’t really contribute much value. Whether you want to see how bad your SAN sucks, or are seeking to collect fact based info you can use in your next vendor negotiation, VI is the stuff.
7. We need a damned ecosystem of vendors that plug and play with a Web Services management bus. (We could also use the bus itself.) 2009 will see some heavy development effort in this space beginning with a C-4 Summit that we’ve been developing over the past year, a C-4 Community, and a C-4 Microfactory where vendors can list their wares once certified for interoperability with a Web Services-based reference architecture model. In short, I’m out to boil the ocean again and everyone’s invited. Such an inclusive and standards-based architecture would deliver more value at a lower cost than stovepipe arrays. Vendors of those products need to go the way of GM, Ford and Chrysler this year (whether chastised or chapter 11′ed) for selling their black 4-wheel drive SUVs in South Florida (where the highest mountain is 30 feet above sea level) instead of sensible and greener vehicles we can afford to buy and drive. Don’t get me wrong, I like muscle cars as much as the next guy, but I just need clean, reliable and affordable transportation that gets me from point A to point B. I don’t measure my worth by the box of tin, aluminum and plastic that sits in my driveway anymore than I measure a company’s “enterprise class” infrastructure by how many DMX arrays they have deployed. Generally speaking, I think less of the guy driving the H2 Humvee (the Chevy Tahoe is the exact same vehicle with a different body style and a $40K lower sticker price) when he pulls up beside me in traffic. I think he’s probably got some issues. And frankly, I have generally the same view of the CIO who gushes over how much big iron stovepipe arrays he has deployed on his raised floor. Everyone is selling a box of hard disks at the end of the day. Show me the guy who knows how to manage heterogeneity and can measure service level growth without labor cost increases. He or she gets my respect.
8. Every big vendor seems ready to announce thin provisioning on their array this year. I guess 3PAR and Compellent scared the crap out of them by taking some of their customers’ money. I have a generally negative view of TP on the array. If you have to do it, do it right. Virtualize storage using DataCore Software. They invented storage thin provisioning and are doing it at the network layer instead of down on the box. Their great financial success in 2008 is testimony to the wit, wisdom and tenacity of George, Ziya and Bettye (and their krewe) who told the trade press and VC guys to go chase themselves when everyone saw the big iron guys as the big winners of the storage virtualization wars. Their story is becoming a legend as their success, like Xiotech’s continues to climb. I have yet to find an unhappy customer using their stuff.
9. Related to Number 8, I just saw an email from a trade press pub that quoted a fellow who said loyalty is more important than satisfaction from EMC’s standpoint. It has been a focus on cultivating loyalty rather than measuring customer satisfaction that has made Hopkinton successful at its game. How is loyalty cultivated if not by satisfying consumer needs? This sounds like the BS that has helped tank US financial houses, automotive manufacturers, and so many other industries. Seimens, over in Europe, cultivated loyalty by paying off governments and customers with a huge slush fund according to an article in the Economist last week. Now they are on the hook for nearly a billion in fines and taxes. There is a point where loyalty (purchased via bribes or other incentives) goes head to head with reality. I hope that 2009 sees the tide begin to turn. I will keep calling the vendors on the BS in the coming year.
10. It is with some dismay that I must report that TechTarget is scaling back my speaking gigs this year. Events need sponsors and sponsors are on hard times, like most of us. I will do what I can to keep my outreach to you going, both here and in video blogs that we will shortly be doing with StorageTV. I also want everyone to monitor the C-4 Community when I have it up and running. We need consumers to tell us what’s wrong and what’s right with their infrastructure and even stovepipe vendors are invited to participate to give us their version of the financial and technical attributes of doing things via a one-stop-shop. Maybe with a lot of participation from consumers, vendors and integrators, we can begin right-sizing infrastructure, making it manageable, and moving data intelligently across it.
Goodbye 2008. Welcome 2009: the year that will test all of our mettle.













