Consolidation

From AgriLife WIki

Since we are looking at consolidating the GroupWise post office located off campus, I thought it might be nice to see what some of us see as the Pros and Cons of doing that. I have started a list and hope that you will add to it.

Pros:

• Fewer post offices to upgrade and patch
• Fewer MTAs to monitor, upgrade, patch
• May be able to eliminate some of the Secondary MTAs
• Fewer post offices to monitor
• Fewer points of failure
• Post offices physically located closer to those doing primary management in case of hardware problems
• Specialists moving from one center to another won't need to have their mail moved
• A single contact point for GW user support could be created. GW users would pretty much need to call EIT on Campus since all the resources are here.

Cons:

• Slower transfer time from the workstation to the server (may be noticeable with large attachments)
• No access to post office if WAN is down.
  • Currently they can compose and send messages that will be delivered when WAN is fixed - could not do that after consolidation
  • Currently they can read messages received before WAN failed - could not do that after consolidation
• More difficult to clean up - Currently I can run an audit report on a post office and send a list of inactive users to one or two people at the center who can tell me the status of those users. After consolidation an audit report on the post office may need to be sent to 24 people at 14 centers to try to find the status of those users. Many of those people probably wouldn't look through a list of 90 users like I sent to Weslaco this morning to see if they might recognize one or two names.
• More difficult to manage distribution lists - Currently, all users are placed in a PO distribution list which can also serve as a center distribution list. It is easy to tell which ESS or EPS list they should be in. After consolidation, I suspect center distribution lists will be necessary in addition to our current lists. It may be more difficult to make sure that people are in the correct EPS or ESS lists.
• From a hardware viewpoint, the more accounts we have on a server, the more folks that are down when it is offline. This can be mitigated some by using the best hardware we can and using several consolidation servers.
• In the event of a failure that requires restoring a backup it will take longer since the post offices will be larger.
• Increased traffic to a single location, here on campus, could become problem.

Other:

• One item that I have heard listed as a Pro is easier user administration, but in my opinion, user administration is easier when the users are in a post office that is in our primary NDS tree. That would happen if we consolidated, but it would also happen if we did not consolidate, but we moved the center servers into the Ag Program tree, which is also being discussed.
• I have also heard the argument that using caching or remote mail will solve some of the cons, in particular, the WAN going down. I agree that if the WAN goes down while a caching user is working, he will be able to continue to read his mail, compose messages, etc. and if it comes back up while he is still working, he may not notice that anything happened. However, if the WAN is down before he starts up GroupWise and/or the WAN is down when he gets finished, the cons listed above will exist. He will not be able to see messages on the server before the WAN went down and messages that he composes will not be sent even when the WAN comes back up if his GroupWise is no longer running.
• <by Mike> One "con" is that the caching client becomes a "requirement" after a certain number of users become "active" on a consolidated server. We can overcome this issue by using several consolidation servers and spread users across them. That way, those that prefer the on-line mode can use it.
• Another "con" is the hardware. When we put a lot of users on one server, we need to setup a stand-by system to switch "on" automatically if the first goes down. This means redundant CPUs and data systems. We've not had the resources (money) to setup such a system in the past and if we did not and the system was down, we would inconvenience a lot of people.

Outside the BOX

After reading this wiki and talking with the Robert and Jim, I am not sure there are any true advantages of only consolidating the GW post offices. There is no savings in equipment in the field because there will still be 2 servers at each location, no savings in scale as far as increasing the service we provide (GWIM), does not fix the tree issue, and there are no savings in time management.

However, I am not sure any of the cons hold up either. Agents is a great example. It is rarely down, few if any agents complain about slow transfer rates, and we seem to manage it with dropping users and adding users. Plus if you combined all the non-agents in the West Region, I don’t think there would be as many users on the agents server, so I doubt load would be a problem.

I think we need to think in bigger terms as far as consolidation. Do we need servers at all Urban, District office, and Research offices? Why not collapse all server services (web servers, iFolder services, Novell Shared drives, GroupWise, etc other services) down into regional servers located around the state? At these 4 to 6 locations, then invest in more infrastructure like bandwidth for these regional EIT hubs instead of computers that are depreciating in value and technology at a very rapid rate all across the state. This model would seem to “fix” a lot of the issues we have with NDS, Tree, and the number of server backup problems, and monitoring. For example, D6 has a grand total of 11 users yet we spend the same amount of money on equipment and support, and time setting up and managing services, as we do in at other locations that have 4 or 5 times as many users. Another examples would be El Paso Research and 4H centers. I think there about 10-15 users at each of these sites.

So, in short, I think Consolidating only GW post offices really is pretty futile and most likely will not save us any money or time, unless that is the only service we are providing at a location. And by consolidating, we can reduce or eliminate equipment as we abandoned the site.

Another idea would be to abandon the location model completely and start over. With regional servers and local file servers at each location but all in the same tree. All services except file sharing would be provided at the regional level. Then back the file servers up with r- sync type technologies to regional sites each day.