Successful installation of network components is no accident. It happens by choosing the best equipment for the job and by preparing for successful installation, operation, and maintenance. Plug-and-play sometimes means “plug it in and it immediately plays” and sometimes means “plug it in and play and play until it works.” The truth is, as always, somewhere in between. By reviewing the following checklist, you can be assured that you have done as much as you can to ensure the success of your server storage installation.
- Configuration–Choose the configuration that meets your performance, storage, availability, and serviceability needs by consulting with your sales engineer. Stick with standard configurations for the best price/performance and ease of service, unless your requirements dictate otherwise. Ask the difficult questions before buying anything. The higher your availability requirements, the more redundancy and component removability you require. Consider redundancy for disk channels, host buses, interface cards, and even servers, in addition to the disk drives and internal disk array components.
- Backup–Ensure that your backup device and procedures are adequate to handle the increased storage in the time available.
- Power–Ensure clean power and UPS protection. Add a cache battery backup option for added data protection.
- Interference–Check for nearby sources of electromagnetic interference, such as banks of modems on Web servers.
- Load–Is the system totally loaded already? Will the addition of one more device, especially a high-performance RAID array, push the load limit over the edge?
- SCSI interface–Is the interface fast enough to avoid bottlenecking the new high-performance equipment?
- Service–Decide who will perform remedial service. Do you need 24×7 service availability, or on-site third-party service? Will you train your people?
- Key phone numbers–Post 800 hotline numbers in conspicuous places. Post beeper number (and perhaps home phone numbers) of key internal personnel. Prepare pocket summaries with this information for key people to keep handy.
- Spare parts–Are advance replacement parts that are supplied overnight sufficient, or do you need on-site spare parts?
- Remote alarms–Would you benefit if your system were programmed to cry for help by automatically dialing designated pagers with an optional alarm?
- Remote diagnostics–With an optional modem, you can enable your system to receive remote diagnostic calls. Temporary system passwords can be provided for use only when the service is needed.
- Performance benchmark–Naturally, you will be curious to test how much faster everything runs with new high-performance storage. Time your longest batch jobs and measure client response to lengthy transactions before and after installation.
- Capacity assessment–Based upon performance benchmarks, you can make a reasonable estimate of how many users you can support on the existing server before you need to add another one.
- Scheduled downtime–Make sure there will be enough time to properly install the system. Allow plenty of time and anticipate that it will take longer than expected–Murphy is ever present.
- Test period–The more critical the application, the longer the shake-out period should be. Several days to several weeks of running diagnostics, exercisers, and representative applications is a reasonable precaution to take before committing your entire enterprise to a new piece of equipment–including RAID arrays.
- Training–Make sure one or (preferably) two or more people are trained on the system. Training includes setting up RAID arrays, swapping components, rebuilding arrays, simulating failures and, most importantly, practicing what to do in the event of actual failures.
- Data protection–RAID is no different than any other storage device when it comes to protection from viruses and accidental or deliberate deletions. Implement the same file-protection and record-locking strategies you would use on a non-RAID system.
- Data recovery–Database and transaction processing systems are often implemented on RAID systems. If yesterday’s backup is too old to be a useful data recovery system, make sure you have the ability to roll the database forward in time with journal files or similar capabilities. Store journal files on a separate physical device from your primary database. No system is perfect, not even RAID systems. You still need to protect your data as you would with any storage device.
- Data backup–Typically, RAID systems significantly increase your total storage capacity. Make sure you do a complete backup frequently and incremental backups at least daily. Test data restoration periodically to ensure that you remember how to do it and to ensure proper operation of the tape system.
- Data security–Make extra backups and keep copies off site.
- Upgrades–Certainly there are other improvements on your agenda. You should implement one major change at a time, so that problems are easy to identify and correct if you are upgrading an existing server. You wouldn’t want to upgrade the operating system and the application while also installing new storage and backup systems. If you need to make all these changes and want to avoid sequential disruptions, build an entire new server and then test it intensively before deploying. Keep the old server as a backup until the new one is proven.
- Communication–Stay in touch with your service provider. If you keep this person informed, he’ll be familiar with your installation and your capabilities and will be better prepared when you need help.