Downtime. A good thing you would think. Time to relax, put your feet up, enjoy yourself. But not if you own and run IT systems. We now all expect everything to be available all the time. Yet computer hardware does everything it can to make things difficult. Especially disk drives which have moving parts, so inevitably break. Systems become obsolete and need changing for new shiny ones. And software needs constant attention to keep things running the way they should.
There are two types of IT downtime. Planned and Unplanned. We have done many things over that past 20 years in the storage world to deal with unplanned downtime. A long way from the only answer being to ‘switch it off and on again’. RAID schemes to protect against disk failure, Snapshots to protect against corruption (data, not the mafia kind) and Replication and Vaulting to protect against bigger disasters. However, although systems are more reliable, they are also way more complex. So, little has been gained in the fight against planned downtime - the time needed to migrate data every few years onto those new shiny hardware systems and more regular software patching and upgrades.
Data migration between old and new storage systems is a costly, time consuming exercise. Even more so given that the amount of data organisations need to keep stubbornly refuses to stop growing exponentially. Most people keep their storage systems for 3 to 5 years. And so, write off the value of them over that time. It’s not uncommon for it to take 3-6 months to do that initial data migration. And of course you need the old system while you do this. So, the actual useful, production life of a system can be up to a year less than you’re paying for – 6 months to migrate data onto it, 6 months to migrate off onto the new system. Some estimate that migration accounts for up to 40% of the Total Cost of Ownership (TCO) of a storage system. And it’s a pain in the *!%# for everyone involved.
Regular planned downtime for maintenance and upgrades is equally annoying. These days each storage array typically hosts at least 10 business applications. To make a change to that array – an upgrade, replacement, repair – means taking all 10 applications off line. This is traditionally done at the weekend, during the night when it has the least user impact. Not the best way to spend your Friday or Saturday night, but there is also another problem. While the actual outage time might be 2 hours during that ‘planned downtime’ window, it can take 38 hours (a recent estimate from NetApp IT) to do the planning, negotiation and scheduling for that outage – 16x the time to actually execute the change! And that’s just one repair for one system. Multiply that by the number of arrays you own and it gets to be a real problem. If you have 150 arrays, it equates to employing 3 full time people just to manage this process! And the 24 x 7 working global organisations now expect means any downtime can have a serious revenue impact on top of this.
All of this is very depressing. And an inevitable fact of life for the lucky storage array owner? Well, it was until recently. NetApp’s clustered Data ONTAP storage software can now deliver nondisruptive storage operations (NDO). Removing the need for you to plan downtime for migration or maintenance. In plain English - NO MORE PLANNED DOWNTIME for your storage infrastructure. Might sounds impossible, but customers around the world like EMBL, Plazamedia, ThinkMoney, VGTRK, Photobucket, Turner Broadcasting, CERN and many more are already enjoying the benefits. This week we released the latest version, Data ONTAP 8.2.
To understand how this seemingly impossible NDO is possible, join me and a few friends for a webcast at the end of the month. To register, click here.