First I have to gush a little about the PASS Summit. Skip ahead if you like - this type of stuff doesn’t make for the best reading, and I won’t be offended. The Summit was amazing as ever this year. The thing I love about it, in addition to just the wonderful, supportive atmosphere, is that most of the people involved are real professionals doing real work, on the ground, with the Microsoft SQL Server stack. There’s some marketing spin, of course, but it’s tempered. You’re as likely to hear someone say “that doesn’t work” or “that’s still vaporware” as “this is the marketecture/kool-aid/pitch/roadmap for all things SQL Server.” I loved the content and the chance to meet old friends and new from the Intertubez.
I generally use the Summit as a chance to pick my head up out of the day-to-day details and pressures of work, and try to look ahead. I went to an advanced driving course here in Seattle at Pacific Raceways a few weeks ago (thank you Kimberly for giving me a push to do that!) and one of the primary exercises they drilled into us students was to look ahead. Look ahead, through the curve, anticipate what’s coming, because at 120 it’ll be here in a second and you’d better be ready.
The second concept was that you drive a car fast with smooth, fluid motions. You don’t jerk the wheel or jam the brakes; twitchy is fundamentally bad. The way to be fast has everything to do with planning ahead, choosing a line, then executing with the minimum inputs. Don Kitch, the lead instructor said, “lots of people want to see me drive their fancy cars on the track: the key to that is, watch what I don’t do with the car.” What he meant was, it shouldn’t be flashy or exciting to watch him toss the car around. That would be missing the point. I think the same is true, if perhaps at a slower pace, for a DBA. Anticipate what’s coming, steer your organization and your systems with a minimum of drama. Do it quickly. Make it look easy.
So, what’s coming up around the next turn? I think the last couple of years of PASS Summit give us some clues.
You’re Gonna Need Peeps
The first is that all of us are going to need to lean on each other in the SQL Server community. The pace of change and the growth of the SQL Server family of products is astounding. More and more we’ll all be under pressure at work to know, or at least know about, the whole stack, and that will become less and less possible. That leads me to what I call “Peep Consulting:” lately, through social media, online contacts and training opportunities I’ve been able to come out of my shell a bit and tap into the community, to real benefit for my work and for my company. Peep Consulting is a way that a bunch of us in the community can collaborate, informally, to make each other look good, and collectively stay ahead of the massive explosion in all the SQL Server products. It’s not about long engagements or big projects, necessarily, it’s more the #sqlhelp tip, or the hallway conversation, or the email list that can steer you in the right direction, from people with more experience in some specific area of the products. People who aren’t doing this type of networking will be at a great disadvantage in the coming years, I think, as the development of the products accelerates.
Things will get Simpler, More Complicated, Slower and Faster
This notion is a bit hard to express, but the trend I see looks like this: the environments we work in, especially where I am in the small/medium business arena, are going to become ever more sprawly. The multiplication of apps and VM’s (Windows is practically the new App, as people spin up a new VM for just about every single service or application) will cause a multiplication of virtual servers, instances and databases. At the same time, the tools for administering that sprawl should improve, so it won’t be quite as daunting as it was two or three years ago. I think we’ll find we are faced with a much more complicated, more heterogeneous environment, with more different products, from the relational engine to SharePoint/PowerPivot and everything else. But we’ll also gain tools to grapple with that more effectively. Sadly, the idea that we’ll ever have a single data source for a small business – one database to store the company’s data in a consistent schema -- seems truly dead. It would be useful, but it doesn’t seem possible. In fact we are headed the other direction, rapidly.
Second, I think the hybrid of cloud services like Azure, infrastructure as a service, and so on, will certainly change the landscape of small and medium business. In a few years a small company that has it’s own servers in it’s own data center or server closet may be a real anachronism. Smart companies will get out of the business of buying servers and electricity, diesel fuel and air conditioning, and will leave that to larger commercial data center operations. Ironically, though, that will increase the trend above, where things get even more complicated to plan and operate. The types of decisions will be different (more about software architecture and less about wiring) but no less complex. Instead of the care and feeding of a cold room full of machinery, we will face the care and feeding of an unbelievably complicated collection of cloud services and data, with real integration challenges.
On the slow and fast front, we should see a continuation of what has happened to this point: software gets ever slower and more complicated, and hardware gets faster just fast enough that nobody cares much. We will continue to burden our servers with layer on layer of additional complexity – I’m looking at you, VMs – because it’s expedient, and because there’s no economic advantage to combatting that trend by making more efficient software. It’s always more economical, it seems, to work around a complex problem with a new layer, than to solve the problem. Allan Hirt had 12 VM’s running simultaneously, with simulated shared storage, on a laptop in his precon. Q.E.D.
Bits vs. Strategy
The combination of those trends I think means that as a work-a-day DBA, on the ground, things will pivot in the next couple of years. Classic problems like backup strategy, DR, indexing strategy, query performance and setup of complicated machinery like failover clusters will probably become commoditized. The software is just making that stuff easier with each release, and eventually it’s not going to be such a specialty skill. There will still be the need for some high-end tuning of large, high performance systems, but the bulk of stuff out there should just work without quite so much effort. When a $10k server has a TB of RAM and PCI flash storage on board, and DMV’s mature to render visible all the cryptic performance issues of the past, it just gets easier to run a small system.
On the other hand, the strategic decisions about where to place a company’s data, how to navigate the complexity of a blended system of on-premises data and cloud services or IaaS, which of the multiplying platforms make sense – from Hadoop to SQL Server to DAX or SSAS or PowerPivot – to keep a company agile, these decisions will become the things that are most valuable in small or medium businesses. All businesses will want to move in an agile way as they or their customers demand taking advantage of the multiplying new services out there. Most businesses I imagine will also want out of the datacenter management field as quickly as is practical, and will not want to own hundreds or thousands of servers any longer. More capability, less equipment.
This is where the SMB IT staff or DBA will differentiate him or herself – I think if we just keep our heads down, tuning query plans, that we’ll miss what’s about to happen. If, instead, we can look strategically ahead, and guide organizations into opportunities for increase capability and lower cost, we’ll stay relevant, and also have more fun.
But then, who really can predict the future?