You may find many incorrect statements about IBM XIV on various blogging sites (mainly put up by nervous competitors such as EMC, Compellent, Equalogic, and Left Hand!). Here are just some of them and we "Dispel the Rumour";
1. If you lose a 2nd drive on a separate module, you lose all your data.
Firstly, the likelihood of a double disk drive failure occurring in separate modules in a single XIV is once in every 11,580 years!! The chances of a fire destroying the datacentre are much higher!!
Secondly, IBM XIV’s SMARTS technology on the drives identifies 90+% of drive failures before they occur. When the system gets an indication of a potential problem from SMARTS, it creates a third mirror of the data on a redundant drive and then phases the failing drive out of the system. Data is not exposed during this process, so there is no risk of a double-drive failure.
2. The IBM XIV unit needs to be running with 79TB of usable storage within 12 months of purchase.
This is not true. The minimum purchase is 6 modules or 27TB usable and additional modules can be added as and when required. Also IBM is able to provide customers with an even lower entry point with the XIV Capacity Upgrade on Demand program.
3. Using 180 x 1TB disks to achieve 79TB of usable storage is inefficient and wasteful.
This would be similar to the amount of usable disk storage available in a typical business critical environment running disk storage under RAID 1 or 10 with hot spares.
4. IBM XIV is only scalable to 79TB.
At present this is correct. However, 2TB will shortly be introduced doubling the amount of usable storage. In further releases, it is planned that customers will be able to scale out, connecting multiple XIV systems together. In addition, up to 15 arrays can be managed from a single console.
5. Using SATA drives results in much lower performance.
As data is broken down into 1MB chucks and then spread across all drives, the I/O is comparable to other non-SATA based systems. In addition, performance actually increases as more disks are added due to IBM XIV’s grid architecture.
6. SATA drives are much more likely to fail than FC or SAS disk.
The latest generation of SATA drives are now as reliable as Fibre Channel drives. Also the unique architecture of XIV means that all disks operate at the same speed with no “hot-spots” in the array. If a drives does happen to fail, the resulting rebuild is spread across all other production drives, resulting in rapid rebuild times with very little additional workload being placed on each disk, minimising the threat of a second disk being lost. Rebuilds under RAID 5, where the hot spare has not been in production but is now having a huge increase in the demands placed upon it is increasing the likelihood of a second drive failure.
7. IBM XIV cannot replicate asynchronously.
Asynchronous replication is available for testing purposes now and will available for production environments in the next firmware release, due September 2009.
8. IBM XIV is very expensive.
The price per TB is very competitive against Tier 1 & 2 solutions from other vendors such as EMC and the pricing includes licences for all software functions and updates. Hardware maintenance can also be purchased for the anticipated lifespan of the system and there are no hidden charges further down the line.
Other savings can be achieved from dramatically reduced storage management costs and lower power and cooling costs. TCO studies which take into account full hardware, software and operational costs under true utilisation provide demonstrable results of significant lower ownership costs to that of other comparable solutions.