The feature that I find most valuable, but I don't see used often enough, is really applying the fault tolerance. In the last two companies I've worked with, they always categorized or contained different type of devices. There were routers in one box and firewalls and other switches elsewhere. They really shoot themselves in the foot because you need to be able to understand in the grand universe what's breaking down where and why. Where on the path is the issue occurring?
Even though this seems to become a very strong part of the application, it's not really being applied because people want to categorize and put everything into little boxes. They don't really understand global collections. That's where you would categorize devices by types or something of that nature. That's something I've been trying to promote in our own company to really get a better sense of a more organic view of the network. We can then see if the problem is with a server, a switch in between, a router, or a firewall. We want to know where along the path is the issue occurring.
Improvements to My Organization:
We are using Spectrum as a conduit for other tools. For instance, we are bringing in SystemEDGE to send traps. We are going to use that to replace NSM, which is an old tool that is not supported by CA anymore. This is our replacement for auto-ticketing, and things of that nature.
When I joined the company a few years ago, they were looking for significant ways in which we could reduce incident tickets, severity tickets, with the different severities, to monitor the duration and improve repair time. With the implementation of the “shallow water” approach, we've been able to significantly cut down on the number and duration of outages and the number of severities in all the categories.
The biggest benefit is
really getting away from Java, because of the way Java is a tool that needs to
be updated all the time. We found a lot of issues with users not being able to
log in because the Java version on their PC got updated and is no longer compatible
with Spectrum. Getting away from Java was a nice touch. We'll be integrating
UIM, and we're probably going to be putting PM into Spectrum, as well. We will
be integrating all those into that particular tool. I like that aspect, as well.
Room for Improvement:
They really need to work out some bugs. Getting away from Java is a good step. I'm looking to see how well we can integrate performance manager (PM) and UIM into that, and see how they all play together. I've got a distinct feeling that there's going to be difficulties that are beyond what they're advertising as to how nice and easy it's going to mesh them together. I think there's a lot of work that still needs to be done in that aspect.
I would like to see easier integration and implementation of device MIBs. I usually find that, whenever we go to add a new device, the MIBs haven't been certified yet. This is a painful process to get them working in order to find alerts, associate cost codes with event codes, and things of that nature. If they can make that process easier, that would be fantastic. That's probably the biggest challenge I found; when I'm dealing with event codes and making sure that they are processing things as necessary. We need them to understand what's coming from the end boxes that are throwing the traps.
Stability could be improved. I find on a regular basis that those of us on-call tend to usually have to go in and restart or reboot a Spectrum server because the landscape's gone down. Then we have to worry about rebuilding databases whenever there's patching. This has gotten better with the latest releases, but we've still got to resolve that issue. For quite a few years, we'd have to go and sit there and wait for patching to occur, so we can go and repair all the databases after the fact because Spectrum couldn't shut down the database fast enough before the reboot.
The stability is a lot better than eHealth. I noticed that eHealth was not being talked about a recent CA conference. I like the fact that they doubled the capacity, apparently, and they're going to be doubling that again pretty soon from what I heard. eHealth is obviously being put out to the pasture. I agree with that because it is old technology and needs to go away. This is why my company's actually looking at PM and that kind of stuff to replace it.
I like the scalability for sure. It's definitely helping out because we're growing in a huge way. We could go from 10,000 to 20,000 devices, and then that might even go up to 40,000 devices. This solution might be able to keep pace with the growth that we're seeing. As far as end-user equipment work setup goes, I
don’t think it will be scalable.
I tend to be on call on a rotating basis, so whenever there are issues, I am one of the folks at the repair site on my weekend. I try to fix problems and issues. In terms of CA Tech Support, I'm sure I've got a reputation because I didn't really like it when they decided to push a lot of off-shore tech support on us. I really fought against that, and thought that it was not appropriate for the kind of contract that we had with them. I let them know that I don't care for how often they want to push professional services on us as a “fix it issue”.
I came to this company
when the tool was already in place, but I was involved with the latest upgrade.
I was not a lead engineer, but I offered support. I was involved with the
project that upgraded all the tools.
Other Solutions Considered:
CA knows about the difficulties with eHealth when going through the upgrade process. I pushed myself to look at my company’s other options because across the board, we're looking for some new implementations. I've been looking for eHealth replacements from other vendors even though our company is CA-centric. As an engineer, I can't really narrow the scope so that we only look at what CA gives us because that's really not to the benefit of the company.
I know CA might not like this but I would suggest looking at other vendors in the market and really see what the completion has to offer. CA has got a large footprint and that's fine, but as an engineer, you should always be looking for what's best for the company regardless of the vendor.
Disclosure: I am a real user, and this review is based on my own experience and opinions.