I have identified a new medical condition, 'Speedy Storage Dependancy'. This addiction is suffered from many developers, DBAs, PMs and many others in the IT industry. It can take many forms depending on the addict but the most common form is based on the instant gratification recieved by putting your poorly written and designed application onto fast disk.
Unfortunately, the addiction is often accompanied by another addiction 'Supersized Storage Dependancy'. This has the impact of making the addictions exponentially more expensive as unfortunately no company has yet developed a 'SuperSized Speedy Storage Device'.
Often the excuse for this addiction is that it is too hard to change now and the alternatives are just too expensive to comtemplate!
As the enablers to this addiction, we in the infrastructure teams need to stage an intervention and develop alternative coping mechanisms. Otherwise the dream of Clustered Storage, Cloud Optimised Storage and fluffy Fairyland Storage will just remain that, a dream!! We need to point out the behaviour which leads to these addictions but in order to do this, we need assistance from the manufacturers of the addictive substances.
Unfortunately, the manufactuers of the addictive substances appear to be more interested in creating the crystal meth equivalent to yet further fuel the 'Speedy Storage Dependancy'. Just say No to SSD until a suitable dependancy plan is developed!
We need bring the consequences of this addiction to the attention of the addicts. I am going to campaign for a health warning to be put on all disk arrays in future, 'This Disk Array may lead to overweight, unfit and unsafe applications which may damage the health of your company!'.
SSD = FUD FUD & more FUD 🙁
Trying to generate a band-aid to solve a problem in a place it shouldn’t be fixed…
SSD could have some future use when :-
a) Capacity increases
b) Price drops massively
c) Reliability improves – array based SSDs fail as well you know 🙂
d) Volume managers & File-Systems are adapted to make ‘sensible intelligent use’ of SSD properties
e) Somebody provides me a TCO that proves investing is SSD is better than the alternatives (Host memory, host processor, In memory databases, Array cache, improving the developer/DBA skills etc)
f) The SSD marketeers remember to include the full context when they are buzzword hyping… Yes there is a latency improvement – but in order for the app to see benefit another €300k of change is required on the old server h/ware etc
We should introduce a vendor fine scheme for buzz words, hype and BS… a “3 strikes and you’re out the tender” kind of approach would work for me 🙂
Thank you for this post, this could help a lot. Addiction is one thing that can’t be treated easily.
-mj-