I always enjoy reading posts on my friend and colleague Steve Duplessie’s blog. Todays is exceptional. Steve has the rarest of talents, the ability to cut through the noise and get straight to the real issue. That’s why Enterprise Strategy Group is the best analyst firm in the world and has a client list that reads like a who’s who of IT. It seems that if you are not an ESG client you aren’t going to cut it. Ask around, check it out I am right.
So back to Steve’s blog. Solid State Disks (SSD) are the next hottest trend after cloud computing and just about everyone is talking about them. SSDs are the death of the spinning disk. SSDs will solve Global Warming. SSDs deliver incredible performance so we can stop thinking about hotspots on storage arrays. Well SSDs are cool they are disruptive, (some of you might know that I am on the Fusion-io Industry Advisory Board so I absolutely get it) but they don’t solve world hunger.
In fact as Steve says in his blog all that SSDs do is fix an element of a problem. A highly technical problem that deals with and avoids stuff like short stroke disks and wide striping. But actually that just isn’t that relevant. Nor is Cloud Computing, or Blade Servers or air side economizers or any of the myriad of technical stuff that we hear about every day in our media.
In a perfect and simpler world, everything would be joined up, compatible and planned. With distributed systems everything is done in silos that are absolutely not joined up so things like balancing workload, capacity planning and management are alien concepts and counter intuitive to the modern operating model. Point solutions, low hanging fruit projects executed in an environment where enterprise architecture is so divorced from the customer and IT operations that it is meaningless. It follows the teenage mindset that putting a sports exhaust on a rusty, beaten up ride fixes everything.
We need a framework to operate in. One that helps us to continuously migrate towards a better IT. One that says SSDs are cool and if we use them in a thought out way, they fit the framework and we move forwards and it is good. Otherwise all the great stuff is just another loud chrome plated exhaust system on a teenagers pride and joy.
Comment by Sjoerd van Groning on 31 May 2009:
Replacing the exhaust might not fix the whole car. But if you have a sports car with wooden 18th century tires, putting some good racing tires on it might actually help.
I think key to performance improvement is profiling the whole path from user to data. And as we have allways done, add cache when needed. cpu cache, L1, L2, L3, cache data in memory, disk cache. This actually does help. Disk io is for a long while a bottleneck in performance of servers, especially database servers. I think the hype is partly correct, though it doesn't fix global warming for mass storage.
Haven't done any math on it, but a 2TB rotating disk might be more efficient energywise at the moment. But if you are io-bound, use SSD if that reduces the number of servers. replacing a 40 disk raid with 1 SSD disk might acyually help. Do I think SSD is really helping but to the math, measure it and invest wisely.
Now let's find the next bottleneck.
Comment by thehotaisle on 1 June 2009:
Tim,
I think you are right, this is Enterprise Architecture – however there are a number of critically important problems with how we do Enterprise Architecture today. The first is that it is often done in a vacuum without involvement of the business and customers. As a result the output is flimsy and badly supported. Architectural decisions become overturned for short term expediency reasons and this causes enormous hidden structural costs further downstream. Secondly Architectures are often built as if changes never happen but actually IT systems are in a constant state of flux and change.
Modern Enterprise Architectures must be flexible and have change built in structurally, being able to absorb the latest technology and thinking without throwing away the old stuff.
I sense we are some way away from that thinking being widely accepted.
Steve
Comment by thehotaisle on 1 June 2009:
Hi Sjoerd,
Thanks for your comments.
Here is a radical thought for you! What about thinking that just going faster is not the only issue that Operations professionals have to contend with?
What about thinking about maintenance and support? Keeping this simple, standard and supportable often has a better business outcome that just investing the the latest go-faster stuff. That is really the message I was trying to get across.
Steve
Pingback by Teenage Mindset in the Data Centre « Enterprise Strategy Group on 28 December 2009:
[...] Read the entire blog entry here >> All views and opinions expressed in ESG blog posts are intended to be those of the post's author and do not necessarily reflect the views of Enterprise Strategy Group, Inc., or its clients. ESG bloggers do not and will not engage in any form of paid-for blogging. Click to see our complete Disclosure Policy. For important information about using this content, please review our Terms & Conditions Tags: blade servers, economizer, IT Operations, Servers, Steve Duplessie, Storage [...]