Sean Maxwell –
If you work in an enterprise IT organization, and you’ve been running mission-critical applications supporting your business for the past decade, odds are that your IT infrastructure is undergoing major stages of modernization. Through the advances of virtualization, cloud, and now all the chatter about the software-defined data center, the challenge to deliver the same level of performance and availability is greater than ever
Perhaps the most visible change that customers experience involves the transition of legacy open systems from physical to virtual to cloud environments. I’ll be the ‘master of the obvious’ here by stating that this transformation is not only time consuming and costly, but it also introduces a tremendous amount of risk to your availability. This change also creates obscurity throughout the data center to the performance of your mission-critical applications. Why is that? Because while virtualization and cloud can simplify IT operations and save a significant amount of money, it also introduces a painfully unintended consequence: you lose 100% of the real-time visibility to your application’s I/O.
Think about it…in a legacy world where one application rode on one physical host connected to one physical storage array (or LUN), life was pretty good. If a business unit or application owner called to complain about their slow application, or worst case, they experienced hard downtime; at least you knew where to start looking to find the culprit. Plus, you probably owned every vendor-specific, device-oriented tool (Host, HBA, Fabric, and Storage) ever sold. While none of these tools gave you systemic, real-time information about how the infrastructure was delivering on the service you were chartered to provide, at least you could cobble together a decent picture of what went wrong after a few days (or weeks), solve the mystery and get back to work.
Virtualization is now pervasive, and at every layer of the IT Open systems stack. It’s marvelously, mind-blowingly cool stuff (technical term). I’m just not sure whether we’re supposed to cheer about it, or if should we jeer about it at this point. Let me make a (tongue-in cheek) case in point below about virtualization at the host, SAN and storage layer…
For our friend Mr. Server Virtualization, we now have many flavors of host virtualization for every x86 environment. Not to mention the ‘original’ server virtualization for big iron on AIX with LPARs. Sure, it was nifty when we virtualized the host in the Lab years ago. It even worked like a charm as we rolled it out on some low-level production apps. We marveled at our operational efficiency, achieving 10:1 consolidation ratios. But because no good deed goes unpunished, someone ‘higher-up’ decided that we should now run Tier-1 apps like email, ERP, and VDI on virtualized x86. Oh yeah, and you need to deliver a 50:1 consolidation ratio to make the ROI work out just right. Gulp….
Now let me introduce Mr. SAN Virtualization. 20 years post-inception, Mr. SAN is still the most mission-critical network in the enterprise. Mr. SAN is the central nervous system for the data center, and every FC attached application has its mission-critical I/O traverse that SAN thousands to millions of times per day. But did you also know that Mr. SAN is the most expensive network in the enterprise, port for port? And unfortunately, Mr. SAN is also the most under-utilized network in the enterprise too. Interesting combination, huh? So if it really is so under-utilized, then why is always getting blamed for poor performance, you may ask? Forget for a moment that it’s a black-box device with little to no performance or availability tools (more on that later). When the original Fibre Channel standard was ratified in 1994, and the first four servers were connected via a switch to a single storage array, Mr. SAN was deemed a big success. At that time, no one could imagine how successful Mr. SAN would truly become. Back then, a ‘1,000 port SAN’ was unimaginable. Today, there are thousands of companies with 10,000+ port SANs. And there are dozens of companies with 100,000+ port SANs. No one ever saw that coming.
Hey, I’ve got a good idea since Mr. SAN is getting so big: LET’S VIRTUALIZE IT! Now we get the complexity of routing thousands of virtualized hosts, which are creating millions or billions of I/O requests. The protocol and standards are still the same, as are these same 20 year-old highways. We’ve just increased the speed limits. With this said, it should come as no surprise that virtualization has become the single biggest threat to SAN availability and performance today. Thank goodness we buy everything in pairs.
And last to our Virtualization party is Mr. Big Storage. Mr. Big Storage is a little out-of-control, sort of like a hyper-growth teenager. He makes lots of noise and can’t help breaking things. He doesn’t mean to, he just can’t always control his growth or his impulses. To make matters worse, it seems like everything in his world is changing every day; new media types, de-dupe, thin-provisioning, back-up, etc. So in this ever-changing world, it seems as though we really need something to simplify all this growth and complexity. Wait for it…yes, I’ve got it. LET’S VIRUALIZE STORAGE!! That’s right; you can finally get storage virtualization from every storage company in the business. AND, each of them will pinkie-swear that they’ll virtualize everyone else’s storage in addition to their own. Phew, I was worried about that last part, thanks for clarifying!
While I’m poking fun at each of these technologies, I’m also guessing that you can identify with some of the challenges that I’m referring to. I talk to the largest companies on the planet every day, ranging from global banking and insurance to manufacturing and telecommunications. And you know what? Every organization I talk to describes how hard it is to guarantee performance and availability to their customers. They all share with me how difficult their jobs have become in the past decade. This is due in large part to the unintended by-product of what virtualization has enabled. Am I picking on virtualization, or picking on the companies that supply you the server, SAN or storage? Never! This is a brave new world we all live in. And virtualization is at the center of unlocking business agility and growth, and it’s only getting warmed up.
Stay tuned for part two of my blog where I talk about how customers can solve the challenges that virtualization introduces.
Log in or register to post comments