Ken Brill, the Man Who Defined the Data Center, Dies

dcblogs writes "The founder of the UpTime Institute, Kenneth G. Brill, 69, died Tuesday, the institute's parent company announced. Brill, an electrical engineer by training, is credited with playing an enormous role in shaping the modern data center industry. 'He singled-handedly crafted an industry out of nothing,' said Mike Manos, the chief technology officer at AOL, who had known Brill since the late 1990s. Until Brill's efforts, enterprises had been defining and measuring data centers in their own way, said Manos. 'There was no commonality.' Today, 'you can't go anywhere in the world without people talking about tier 1, tier 2, tier 3 data centers — it's that fundamental,' he said. In 2011, following Amazon's prolong outage, Brill warned that the perceived reliability of large cloud providers was going to lead to problems. 'There will always be an advocate for how it can be done cheaper, [but] if you haven't had a failure for five years — who is the advocate for reliability?' said Brill. 'My prediction is that in the years ahead, we will see more failures than we have been seeing, because people have forgotten what we had to do to get to where we are.'"

Comments

reliability

[phantomfive]

'My prediction is that in the years ahead, we will see more failures than we have been seeing, because people have forgotten what we had to do to get to where we are.'

And considering the cloud isn't exactly known for reliability right now, yet another reason not to trust your data out there.

Re:reliability

[hawguy]

'My prediction is that in the years ahead, we will see more failures than we have been seeing, because people have forgotten what we had to do to get to where we are.'

And considering the cloud isn't exactly known for reliability right now, yet another reason not to trust your data out there.

Is any multi-region cloud provider less reliable than any single-site datacenter? There will always be unexpected disasters (or at least unplanned for, you may expect an asteroid to hit the planet every 50,000 years, but that doesn't mean that you've built the datacenter to survive it) and human error (like "oops, I wish I hadn't dropped my wrench into that panel, 480VAC makes a lot of sparks... it sure is dark in here now"), so it's not clear that cloud providers are siginificantly worse in that regard.

Of course, when a business's primary data center goes down, few people hear about it, while when Amazon has a regional outage (or even in a single Availability Zone), everyone hears about it because it affects hundreds or thousands of companies.

Re:reliability

[pspahn]

But is it only about reliability?

Dealing, in the last week, with a 30 user Exchange outage (MB blew a capacitor, HDDs needed a restore) installed on-site, it made me realize why I originally chose to offsite the new domain's email instead of hosting it locally. The MS shop guys had a different plan and moved it all over to the Exchange server.

So it's now been like three days while they wait for a MB replacement when there would have been nearly no downtime had we been on the service I originally set up.

Reliability doesn't matter when you still have to wait a few days for parts (yes, this happens). Meanwhile you have some MS shop dictating things when a proper cloud service option is clearly the smarter deal.

There's certainly something to be said for hosting locally (or at least keeping a copy), but for most businesses that don't want to deal with some random employee being "the IT guy", offloading this to some facility somewhere for $70/mo simply makes sense. You think Cathy the checkout girl (who took an IT class in college) wants to wake up at 2am to deal with a blown capacitor?

Re:reliability

[aaarrrgggh]

Outsourcing 30 users is likely the top end of a sweet-spot though. Once you hit 50, those initial economies/efficiencies go away, as you would want In-house redundancy, data discovery, proper archiving, etc., which add up to real money in a cloud solution. There are other break-points as you go up depending on organization needs.

Re:reliability

[phantomfive]

The MS shop guys had a different plan.......So it's now been like three days while they wait

They clearly didn't have much of a plan.

It would only be fitting ...

[PPH]

... if he is laid to rest below a couple of raised floor panels.

Re:I thought Google defined the data-center.

[ackthpt]

I thought Google defined the data-center.

http://infolab.stanford.edu/pub/voy/museum/pictures/display/0-4-Google.htm

They just re-defined it. Don't forget, Google began with a ton of commodity PC motherboards stuck in racks.

Now they have money, the redefined their needs.

Meanwhile ... we started with a broom closet and still need to get the water sprinklers removed :(

Re:people talk about "tier X" datacenters, but dif

[CyprusBlue113]

on what each tier means.

This is a case where there are too many different standards, all using the same terms.

In any case, administrative processes are going to affect uptime numbers far more than simple infrastructure redundancy

remember that Google (who has some of the best uptime around) doesn't bother with dual power for it's servers, so it could not be more than a lowly tier 2 datacenter per some standards.

The actual standards define availability redundancy and concurrency of systems, not of individual devices. When your systems are composed of multiple independent devices, it affects what is looked at accordingly.

True in many things...

[CptNerd]

'My prediction is that in the years ahead, we will see more failures than we have been seeing, because people have forgotten what we had to do to get to where we are.'

There are many aspects of our society and world for which this is true, not just data centers.