UTS tackles automated server provisioning

The University of Technology Sydney (UTS) has created a proof-of-concept to reduce the time it takes to provision new virtual servers from a fortnight to about three hours.

The university is one of a number of large Australian organisations to adopt VMware’s vRealize Suite, which contains tools for automating, orchestrating and managing virtualised and cloud-based infrastructure.

vRealize has found wide usage in top-tier enterprise environments, including the Reserve Bank of Australia, Westpac, Woolworths and the Department of Social Services (DSS).

UTS will likely join them in production early next year.

UTS technical services manager Steve McEwan told iTnews the university wanted to cut the time it took to stand up new virtual servers.

“Part of our challenge is that a simple Windows or Red Hat virtual server in VMware can take about two weeks to provision,” McEwan said.

“It goes through about four or five different teams that report into me, the ticket bounces around different people and they do their bit and then it comes back.

“The outcome is that two weeks later, someone gets a very vanilla server or servers, and that’s a really long timeframe to be waiting if you’re trying to do development work or to just prove a concept.”

McEwan said IT was also running the risk of users turning to public cloud to avoid the long provisioning wait times.

The university has therefore built a proof-of-concept where a single server or stack of servers can be automatically provisioned.

Users will be able to go into an existing service catalogue – powered by ServiceNow – used to procure other IT equipment, and choose from about 10 standard server builds.

ServiceNow then makes calls into vRA and VMware NSX to provision the new infrastructure and network firewall rules automatically. It does this using an API developed by Fujitsu and VMware, who also worked on the project.

“We’re not in production yet but we’ve done a proof-of-concept in a dev environment,” McEwan said.

“We estimate that we can provision servers now in about three hours, which is a big difference from two weeks and really a massive improvement from where we were. The aim is to put it into production early next year."

According to McEwan, the ability to order new servers like any other piece of IT was a key part of the project.

“When someone wants to procure a laptop they can order that through the service catalogue, so the thinking was, ‘let’s make that the same experience for servers’,” he said.

“Rather than having a different place to go to provision servers, let’s do that through ServiceNow.”

An added benefit for McEwan’s team is that ServiceNow keeps track of the provisioning workflow.

“The good thing about that is it’s logged properly with a ticket, the change control process is there, and any approvals – depending on how many servers people want to provision – would go to their line manager so there’s workflow in there as well,” he said.

“It’s quite a neat outcome for us because it captures everything, not just provisioning.”

McEwan said the aim was to keep provisioning simple.

“We have around 10 options [for new server builds] in the catalogue,” he said.

“It’s like an 80-20 rule. The [20 percent of] more complicated or unusual requests we would still manage in a more manual way, but if we saw those types of requests more regularly we would obviously add them to the catalogue.”

He expects that automating provisioning of new virtual servers will drive quality and consistency into the process.

“Things like adding [the server] to a patching schedule, making sure the anti-virus is properly configured, whitelisting applications - all of that is handled automatically, whereas before there’d be a couple of manual steps along the way,” McEwan said.