Hop jumps to Pure Storage to avoid big storage complexity

Big suppliers imposing complexity

Until the end of 2017, Hop relied on Dell EMC VNX storage arrays in its two datacentres, linked using a VPlex virtualisation layer.

Kervella vouches for the reliability of the products, but had become very keenly of the specific skills needed to run them. So, somewhat disenchanted, when the Dell EMC kit reached end of life, Hop’s CIO looked to the market for a simpler solution.

“We saw Dell EMC, NetApp and HPE, who all have innovative products in their ranges. But, based on the fact we had previously used complex solutions they didn’t propose these [newer, simpler-to-use products],” says Kervella.

“Dell EMC wanted to sell us VPlex again, while for NetApp nothing was possible without MetroCluster. They didn’t hear what we were saying, that all these components needed specialised skills.”

But time was pressing. Kervella didn’t have time to carry out tests on new solutions, and the suppliers knew that. What the incumbents didn’t anticipate, however, was that people had started to hear about another airline that had a good experience with Pure Storage and that this had reached the ears of Air France Hop too.

“We acted on that experience and opted for two Pure Storage M20 arrays, deployed in active-active mode, at a price that was similar to those of the other suppliers,” says Kervella.

What most seduced Kervella was the easy-to-use approach of Pure Storage. In a traditional storage array, a specialist might take up to three days to configure for a new project. With Pure Storage, a single member of staff at a single interface could build all the virtual machines and their storage in a morning.

“Take the example of active-active functionality,” says Kervella. “Previously, we would have needed to constantly go between EMC and VMware consoles to configure every project. With Pure Storage, space created on the first array is automatically created on the second. Settings are minimal and projects are deployed more rapidly.”

The two M20 arrays were installed in February 2018, with production switched over a month later.

“All our servers are virtual so the migration was easy,” says Kervella. “We used VMware vMotion. We set the hypervisor to execute the VMs on another physical server, which allowed it to migrate very simply to the new array.”

This way of doing things allowed Kervella’s team to migrate all Hop’s applications, from the most mundane to the most critical, in about a week, during working hours and with no impact on users.

Less space, more speed

The chief benefit of the new configuration is physical. Each M20 array has capacity up to 100TB, but in just 4U of rack space. The 60TB of the previous setup took up 39U by the end of its time.

“With drive capacity mostly being flash now, we have noticed that power consumption is about 10% of what it was,” says Kervella.

Of course, flash is also very fast. But for now, the key benefit here is that it allows nightly backups to run 20% to 30% more quickly and to not impinge on the working day.

“The most important improvements are in database dumps, which now execute in 40 minutes where they previously took an hour and a half. However, the advantages of flash will come to the fore in business applications, which will be decisive in our new predictive maintenance project.”

The Air France Hop CIO has discovered that the simplicity in use of the Pure Storage arrays also extends to their maintenance.

“Historically, we have always been dependent on maintenance operations being carried out by the supplier. Typically, controller software updates required one or two people to work on actions specified by Dell EMC. Now the process is totally transparent – Pure Storage handles things remotely and we just get an email to say what’s taken place.”

“With the ease of use and the transparency [with Pure Storage], we have great confidence in our storage. Now we are serene, where once there was stress,” adds Kervella.