This article can also be found in the Premium Editorial Download "Storage magazine: How HP is reloading its storage strategy."
Download it now to read this article plus other related content.
You're on the way to a kickoff meeting for a new CRM project that requires back-end storage which, as the storage administrator, you must supply. You recall conversations with some of the database administrators (DBAs) about the requirements--approximately 2TB--which you feel can be done without breaking your budget. The meeting is attended by several DBAs, the project manger (PM), a few DBA/software contractors representing the vendor and a sales representative you've never seen before.
Before you can say "ambush," the DBAs bombard you with unrealistic disk requirements and configurations, and the PM scratches your name on the Gantt chart, but quickly adds that there are no additional funds in the budget for storage hardware. On your way out, you're reminded that the progress of this project will be monitored at the highest levels in the company.
If this fictional scenario touches a nerve, there's a reason. As technology progresses, storage allocation is becoming less of a mystery. Advances in software allow groups outside the storage team, such as DBAs, to allocate and automate storage. For example, the Automated Storage Management feature in Oracle 10g allows DBAs to allocate storage from a pool of storage at their discretion, and without involving a server or storage resource.
So, do DBAs want to be storage administrators? Probably not. However, as a storage admin, you'll need to decide what DBAs expect from detached storage
To start, DBAs and storage administrators were asked to rate the following criteria on a scale of one to 10, with one being "No concern" and 10 being "Highest concern." The criteria included:
- Backup windows
- Disaster recovery (DR)/recoverability
- Proprietary hardware/software features
- Service level agreements (SLAs) (vendor related)
- Storage tiers
We found DBAs rated their highest concern as uptime, followed by performance, redundancy and DR/recoverability (see "DBAs and storage admins agree ... somewhat," this page). While redundancy increases uptime, it doesn't guarantee it, which is why they're separate issues. Most DBAs would rather have a system up and performing slowly, or operating in a degraded state, than not functioning at all. The average score for each of these four DBA priority issues was eight or above; vendor SLAs, storage tiers and proprietary hardware/software features netted average scores of five or higher (many of the DBAs whose companies had storage teams didn't know what tiering was).
Not surprisingly, storage administrators' responses were similar. Uptime, performance, redundancy and DR/recoverability all scored an eight or above, while proprietary hardware/software features, backup windows and storage tiers received average scores above seven. Scores for SLAs and cost were in the six range. In addition, some storage admins felt scalability was a key factor that should have been included in the list.
The next survey question dealt with disk subsystems and asked: "When selecting a storage solution, what matters most to you?" We used the same one-to-10 scale, with one meaning "Doesn't matter" and 10 indicating "Matters a lot." The criteria evaluated included:
- Drive type, quantity, capacity
- Existing relationships with vendors
- Management direction
- Manufacturer name/reputation
- Manufacturer warranty/service ability
- Past experience with hardware
- Proprietary software for management, reporting, failover, etc.
- RAID types
- Redundant connectivity to the data
- Redundant hardware in the disk subsystem
- System cache
This was first published in July 2005