New Year's Resolutions
(I have no financial interest in any products mentioned in this column except OpenView, which my company resells.)
Its a new year, new century, new millennium! Forget those old standby resolutions like losing weight, learning a new skill or organizing the sock drawer. This year, how about cleaning up your act in the datacenter? Here are a few Im going to shoot for:
The start of a new year even this one! is a great time to address areas that go neglected in the turmoil of daily operations activity.
Archive infrequently accessed files: If a file hasnt been accessed in years, it belongs on tape or near-line media. When you leave old files around you bloat your backup and leave less room on your system for space that can be used productively.
Clean out inactive users: You can take steps to make identification of inactive users easier: e.g., home users to their own group. However you do it, for security as well as management purposes, a periodic removal of inactive users makes sense.
Develop a design for file organization: Rather than have a free-for-all on your system, IT should agree on account/ group/user/structure. Everyone benefits from the ability to find files in a timely manner. And if its done right, appropriate security can be implemented. Also, when everyone buys into the design, it will be easier to locate misplaced files.
Develop a routine maintenance schedule: It is unprofessional, not to mention rude, to interrupt users on little or no notice. Save yourself some grief by settling on a schedule for routine maintenance (patches, hardware and software upgrades, etc.), then share this schedule with all your users.
Keep all critical system materials in a designated area: How low tech can you get? And yet, I have been at countless shops where a simple question like where is the patch tape brings on a look of either terror or bewilderment. Identify a known, secure location and make it a habit to keep all critical materials there. And while youre at it, weed out any obviously worthless materials e.g., old patch, product tapes that might get grabbed instead of the current ones.
Clean up your computer room: The Y2K scare caused many of us to make major overhauls at both the system and data center levels. Others got there from data center consolidations. If you havent made your move yet, perhaps its time to do something about that rats nest under the raised floor. And all the junk piled up, gathering dust, around the console. You know what Im talking about.
As youve heard me state in the past, if it can be automated it should be. Use the momentum of a new year to start your automation initiative.
Implement Job Scheduling: There are several good choices in the HP 3000 market, and one of them is right for you. Regardless of the one you choose, once you have a job scheduler you can begin to create jobs that perform routine system management tasks on a scheduled basis.
Implement Enterprise Management: If your company really wants to go all the way, you cant do much better than OpenView for monitoring the overall health of your network and systems. There is an impressive case study of an HMO that implemented OpenView IT/Operations (IT/O) for their HP 3000s. IT/O monitors all critical system (and network) vital signs and can take automated corrective actions or just notify you decide. IT/O will also monitor trends, allowing you to be proactive. While this is a must in a really big environment, IT/O and its equivalents have been a tough sell (due to up-front cost) in the smaller shops.
Implement Notification: Automated operations are definitely the way to go, if you can convince the money people to make the investment. But your automated operation is not complete unless you have a way to notify appropriate staff when an event occurs. There are several notification solutions for the HP 3000, some of which combine hardware for system monitoring along with the (typically) paging notification. This not necessarily a lights-out solution, either. By using notification hardware/software in conjunction with your automation tools, you can manage by exception. This allows operators to leave the console and only respond when an exception condition occurs, which is a more efficient use of an operators time.
Perform a security audit: VEAudit will check your system for security holes. If you dont have VEAudit, you can get a demo tape from VEsoft. You are almost certain to be surprised by what the audit finds. Take the results to heart.
Eliminate the obvious holes: The biggest problem I encounter at customer sites is too many users with SM capability. And the older the shop, the more SM users there are. Why? Often the explanation is that users (programmers) need SM so they can support the applications. But what is really happening is either (1) the system is a free-for-all, where users can do whatever they think is right; or (2) the account structure has been implemented so badly that SM is needed to operate across account boundaries. (Of course, logging on into multiple accounts is too much trouble.) Unraveling a multiple SM environment is no fun, but unless SM capability is contained, the system will not be stable.
Other obvious holes: users with no passwords; unnecessary accounts (and users), especially with special capabilities; released databases and files; passwords that never change; passwords embedded in job cards; consoles logged on as MANAGER.SYS to name a few.
Implement a security policy and/or security software: System security should be defined in terms of who needs what capability and access. It is possible, albeit awkward, to implement a security policy on an HP 3000 without third-party software. In the case of security the journey developing a security policy is a big part of the reward.
Implement hardware redundancy: In 2000, downtime as a result of disk failure is inexcusable. Unless you can afford the downtime and very few of us can implement RAID or mirroring.
Stay current on OS stability issues: As my experience with labeled tapes which I shared last month showed, it pays to know where the mines in the operating system are. Before making configuration changes, applying patches, upgrading or anything else remotely risky, do your homework to make sure you arent about to repeat someone elses mistake. Sources abound on the Internet, including 3000-L, HPs support site, and 3kworld.com.
Get serious about disaster recovery: A lot of people visited disaster recovery for the first time in conjunction with Y2K preparations. If you dont have a plan, have an old, obsolete one, or did one of those sit on the shelf plans get busy. Remember: Its not too soon to prepare for Y10K.
Dont just sit there, do something! And try to have some fun while doing it.
Scott Hirsh, former chairman of the SIG-SYSMAN Special Interest Group, is a partner at Precision Systems Group, an authorized HP Channel Partner which consults on HP OpenView, Maestro, Sys*Admiral and other general HP 3000 and HP 9000 automation and administration practices.
Copyright The 3000 NewsWire. All rights reserved.