The last month we seem to have spent a lot of time dealing with projects to upgrade various systems. We run a large number of such systems, covering everything from virtual learning environments and wikis to student record systems and timetabling, and also encompassing underlying middleware software. I say "vendor systems" but some of these are open-source projects maintained by a global community. Others are traditional vendors, ranging from large multi-national corporations to niche vendors of HE products.
We have encountered problems with a number of systems, from various sources. Some of these have been bugs in the third-party software. I won't name the systems, nor even say what functions they perform, as I don't want to bring any particular vendor into disrepute. Other issues have turned out to be caused by mistakes in our own systems. In one failed upgrade, we kept the existing system running on one site while we upgraded the system on the backup site, only for the tests to fail; it turned out that the problem was a setting in the load balancer that we were using to separate traffic between the two sites.
Upgrade failures are very frustrating for everyone involved. We have responded quickly in all cases. Usually we have brought an external consultant on site for a couple of days, as a new pair of eyes can help to diagnose problems. The consultants don't always find the underlying problem but sometimes they do, and at least they can confirm that we aren't missing anything obvious.
In most cases we have resolved the issues and the upgrades have either proceeded or are due to proceed shortly. One or two do remain and we are working with the vendors to find a way forward.
We have encountered problems with a number of systems, from various sources. Some of these have been bugs in the third-party software. I won't name the systems, nor even say what functions they perform, as I don't want to bring any particular vendor into disrepute. Other issues have turned out to be caused by mistakes in our own systems. In one failed upgrade, we kept the existing system running on one site while we upgraded the system on the backup site, only for the tests to fail; it turned out that the problem was a setting in the load balancer that we were using to separate traffic between the two sites.
Upgrade failures are very frustrating for everyone involved. We have responded quickly in all cases. Usually we have brought an external consultant on site for a couple of days, as a new pair of eyes can help to diagnose problems. The consultants don't always find the underlying problem but sometimes they do, and at least they can confirm that we aren't missing anything obvious.
In most cases we have resolved the issues and the upgrades have either proceeded or are due to proceed shortly. One or two do remain and we are working with the vendors to find a way forward.
Comments