How much allowance should we make for data expansion

[jrubinstein]

We're investing in a new SAN as part of a virtualisation project. We've been offered a massive range of SAN hardware with varying expansion capabilities and costs. What I want to know is how much should I expect our current data storage to increase in the next 5 years?

 

Obviously as part of the total cost of ownership calculation we need to be clear about when we'd need to grow beyond our current allowance. Most of the proposals have specified (at my request) approximately double our current backup data size. I am assuming that beyond a certain point (say 8 years) there is no reason to expect that any current device would be the ideal way to expand capacity.

 

I'd be very grateful for any advice.

 

Jay

[Geoff]

I think at a minimum you should expect it to double, but somewhere in the x5 to x10 range is more likely over a five year period.

[pcstru]

You shouldn't purchase storage that will serve your needs beyond a couple of years - maximum. Cost per GB halves every few years (drive capacity doubles) so it will be cheaper to expand after two years than it will to provision the space now that will accommodate 5 years of growth. Buying disc space that you won't use for 3+ years is just a waste of money.

[Oaktech]

I appreciate what you are saying, but in our case the way to look at it is to simply buy as much as we can afford, and argue for more than we think we need. If we have to go cap in hand to the purse strings in 2 years time asking to do the excercise all over again our management would point the finger at us and say that we should have anticipated and bought enough to deal with the federation's requirements for the medium term. Management here likes a 'fit and forget' solution.

 

We've seen a doubling in our backed up data size in the last 18 months, with the federation offering more creative courses we expect this to double again in the next 12 months.

[pcstru]

That seems ... bizarre. I submit plans to 'management' that cover a rolling 5 year spending forecast. Sounds like you need to educate your management :-)!

[Geoff]

People problems are not the kind of hardware issues we should be discussing here surely?

[Oaktech]

peoples pereceptions are surely a consideration when deciding how high to pitch your technical proposal?

 

Basically our network gets rebuilt every 5 years, its coming up due and i'm going to have to fight for every penny to do it. Between those times there is no desire to invest in anything other than the bare essentials to keep the thing ticking over as "surely we don't need to spend more, haven't we only just spent xx thousands"

[localzuk]

I budget for a doubling of the amount of data stored every 2 years. So, over a 5 year period that's about a 6x increase.

 

With your backup solution, is this a 'weekly full' type backup? If so, I'd be looking at a d2d or offsite incremental backup now instead.

Edited August 21, 2012 by localzuk

[pcstru]

peoples pereceptions are surely a consideration when deciding how high to pitch your technical proposal?

Yes and no. You need to be prepared to deal with perceptions, you should not pander to them to the point where you allow them to steer you into bad decision making.

Basically our network gets rebuilt every 5 years, its coming up due and i'm going to have to fight for every penny to do it. Between those times there is no desire to invest in anything other than the bare essentials to keep the thing ticking over as "surely we don't need to spend more, haven't we only just spent xx thousands"

I'd do the numbers and show them how and why it is cheaper to keep a network running and accommodating growth by good strategic planning and yearly investment. Technology changes fast enough that generally it's cheaper to buy the capacity when you actually need it than to buy it 3 years before you even start to use it (and add another 2 years before you have used all you planned for). In the 2 year time scale, performance and capacities double while prices half (Ok - not always true but it's a reasonable rule of thumb). The only area where I'm happy with 5 year investment time-scales is cabling and to some extent switches. But most of the cost on cabling is labour and network traffic growth is being driven more by a need to support an increasing number of devices than by the demand from individual devices.