Last week VMWorld took place in Las Vegas, and I was fortunate enough to attend for the second straight year. I love the energy at VMWorld, it is unlike any other conference I attend. The technology, as well as the attendees, appear to be on the cutting edge of enterprise technology. The discussions we have in and around the exhibit hall are worth the price of admission alone. On top of that I am lucky enough to be able to rub elbows with top industry experts and have discussions about the future tech landscape.
During VMWorld there was a hashtag on Twitter, #VMworld3word where attendees would use three words to describe VMWorld. That got me thinking about how I might try to describe VMWorld in three takeaways instead of just words. These three items were common themes in the discussions where I took part, or witnessed, last week.
You can find lots of articles on the internet that summarize all the major announcements at VMWorld. That's not what this blog post is for. No, this blog post is my effort to help you understand what I witnessed as common threads even in regards to the major announcements.
Storage is King, Maybe
Make no mistake about this, everywhere you looked you found someone talking about storage, storage issues, and storage solutions. Flash is the answer for everything, apparently, even if storage isn't your issue. At one point I swear a storage vendor promised me that their all-flash hyper-converged array would cure my polio. The amount of money being invested in storage vendors may be trending downward, but judging by the exhibit hall floor last week at VMWorld the amount of money being spent on storage product development and marketing remains high.
One aspect that these storage vendors seem to either be forgetting, or just not talking about, is the Hybrid IT story. It would seem that the Cloud is a bit of a threat to these vendors, as they are finding it harder to sell their wares to enterprise customers and instead must start focusing on building partnerships with Microsoft and Amazon if their products are to remain relevant. Unfortunately, those Cloud giants rely on commodity hardware, not specialty hardware, which means to me that the storage gravy is just about over. Let's face it, Microsoft isn't about to order a million hyper-converged arrays anytime soon.
The last point I want to make about storage is that their seems to be a mindset that storage is the main bottleneck. Many vendors seem to forget that the network plays an important role in getting data to and from their storage devices. The network seems to be an area where storage vendors just put their hands up and say "that's not us". This is especially true when we talk Hybrid IT as well.
Correlated Monitoring is Lacking
Whenever I had the chance to talk monitoring to vendors and attendees it was clear to me that correlation of metrics and events is something that is lacking in the industry. I believe this to be true for two reasons. First, everyone makes dashboards that show metrics, usually related by resource (disk, CPU, memory, and sometimes basic network stats), and everyone admits that such dashboards are not very good at telling you a root cause. Second, the look on the faces of data professionals when I show them the main virtualization screen for DPA. Once they see that stacked view that allows them to see issues at the storage, host, guest, or database engine layer, the initial reaction was "take my money". It is as if no one on the market is presenting such data in a correlated way.
That's because vendors have spent years building tools that report metrics but do not report meaning. The latest trend now is machine learning and predictive analytics in order to give insight but the reality is you don't need a lot of fancy algorithms behind the scenes in order to do 80% of your work. What you need is for someone to provide you a group of metrics, across your infrastructure, that show the relationships between entities. In other words, is there is an issue with a LUN, can you quickly see what datastores, hosts, VMs, databases, and applications might be affected? For many vendors the answer is "no".
Accidental Cloud DBAs
Given my background and role I naturally gravitated towards conversations with DBAs last week. Storage and correlated monitoring were two of the topics we talked about. The third topic centered around how traditional DBAs today have little to no insight (or knowledge) of how networks work. But with Hybrid IT the reality for most, the data professionals I spoke with last week acknowledged that they needed to know more about networks, network topology, and how to troubleshoot network performance quickly.
Think about this for a moment. When you company starts using cloud resources, and someone calls your desk saying "the app is slow", are you able to quickly look to determine if the issue is related to the network? I tend to do my performance troubleshooting in buckets. It goes like this: either something is in this bucket, or in that bucket. If the app is slow then I want to quickly determine if it is a network issue or not. If it is network, then work with the team(s) that can fix the issue. If it is not network, then I know it is likely something I need to fix as the DBA, and I get to work.
But I certainly don't want to lose hours of my life trying to fix an issue with the application that doesn't exist. The Cloud DBA will also serve as an accidental network administrator as more companies adopt a Hybrid IT strategy.
There you have it, my three takeaways from a fabulous week in Las Vegas. Oh, and here's my #VMWorld3word for you: Fall out boy.