Comment by josefresco | Hacker News Reader

Comment by josefresco | Hacker News Reader

Parent Full thread

"Let's say that again: as of 2013, there were still hundreds of mission-critical IBM internal applications running on physical hardware until I was given the project to transition them to virtual servers."

Is this so shocking or even wrong? I would expect a company with the history and size of IBM to have tons of "old" hardware/processes. In fact it's that history, perceived wisdom and entrenchment that I would assume many in the enterprise world would would see as an asset for IBM. Am I the only one who doesn't think a business should jump on a new tech bandwagon at the expense of a solid and working business model? I'm not saying virtualization or cloud computing is a fad by any measure, but how does a company rationalize rebuilding old tech with the risk that it may be the wrong path/choice or jeopardize a working model?

I know little of IBM and cant say I'm part of the corporate/enterprise market but that fact that IBM has "old iron" in their portfolio seems to be less concerning for their customers and more of a perception issue among the media.

The reason why it wasn't nice was because of the complexity and inefficiency of maintenance. The senior managers who gave me this project wanted to move in that direction in 2010, but couldn't because again, they couldn't get the proper hardware in place in a timely fashion; just delay after delay. Then the Fukushima earthquake happened, which caused an emergency migration of applications from their Tokyo datacenter to their Baulkham Hills datacenter in Australia, which in turn used up all the capacity they were hoping to use for virtualizing everything.

It's not a bad thing if IBM wanted to stay on pure hardware for technical reasons (hey, virtualization can have performance hits, depending on the application). But even IBM didn't want to stay on pure hardware, and that's why it can be considered a sad thing that it took so long.

edit: clarification

Exactly.

Cloud is not the silver bullet. But at the same time, there is not much merit in choosing to be a luddite and not use the capabilities of cloud where there's a genuine need. I feel IBM, to an extent is guilty of not pursuing things on the cloud front and is now paying for it.

Yes, for many customers it's the right solution. In early 2011 we were setting up some IBM technologies on Amazon Web Services for a Houston power company and IBM had just released some EC2 AMIs for Websphere, Portal, etc. If they didn't want to build their own cloud the way HP is trying to, it seems they could've just struck a deal with Amazon to have their own corner of AWS.

"Cloud is not the silver bullet."

You can say that again. And it depends on where you are. In South Florida the cloud - well - sucks. Due to the carriers not being able to keep their networks up. I know companies who moved their PBX into the cloud (I'm going to name carriers) with Earthlink, Windstream, Comcast and ATT - misdirected, noisy (like talking in a wind tunnel), dropped calls; all classic signs of over extended networks. Latency. And you feel helpless cause all you can do is call your carrier and wait on hold for 45 mins.

I know another company, a local gym, which (the owner is a SOB so he deserved everything he got) got talked into moving his cash register into the cloud. I warned him against it. But he accused be of not being "up-scale". His way of telling me I didn't know what I was talking about. Of course, a couple of weeks later his Comcast goes down. He calls me, begging for help. I was like, "Nothing I can do. You have to call Comcast".

In South Florida, the only reliable service is fiber. Even T1's are fragile. And fiber isn't cheap.

You're using a very, very basic definition of cloud- one that really shouldn't exist anymore.

You're talking about moving a physical service into a hosted service; IAAS/SAAS. We're talking about actually developing and coding the IAAS/SAAS solutions to be cloud ready and performant. This infers automation, utilizing version control systems, scaling, stateless servers etc.

What those companies did was move their servers onto the web. It's like me telling you that I'm moving my wordpress blog from my desktop to linode. Just because I'm using linode does not mean I'm utilizing the infrastructure in any sort of "cloud" manner. If the server goes down and I've not designed my stack to be resilient I'm not "cloud." We expected IBM to be designing underlying cloud infrastructure years ago. Instead they bought Softlayer. I'm not sure if they've even done anything remotely interesting since that.

Next year they'll probably start trying to massively hire Openstack developers just like Cisco/etc are doing now (also late to the party).

Semantics, but vastly different.

"You're using a very, very basic definition of cloud- one that really shouldn't exist anymore."

Really? This is exactly what the "cloud" service are selling. Ok there's a middle-guy in there. The actual service provider. But from a customer point of view - this is the cloud.

IBM, Microsoft, Google sales guys are all over the place "selling" the cloud. "Move your infrastructure to our cloud - cheaper and access everywhere on everything. No need to maintain a server, or even a PC. Run your business from your iPad".

Sounds great. And would be great, if we had the communication infrastructure in place to support it. Look, it doesn't matter what's running the cloud if no one buys the "cloud".

Back to your topic.. I agree. But we've seen this story before from IBM (and Cisco and MS). They are late to every party these days.

As for the whole discussion about IBM's involvement with "cloud" - I guess it depends a bit on how you define it. AIX first started virtualization on a large scale years ago, didn't they? And since AIX 5, wasn't every AIX instance an LPAR? PowerVM / VMControl had the essential features of the cloud - virtualization, metering, billing, partition mobility, etc. It was just never marketed in the high volume, low margin way that the public cloud providers are marketing. But, I guess, even then, due to internal politics and protecting their teams, PowerVM and VMControl became like beasts. Software that was produced (E.g. SmartCloud), by edict, had to use it, and those additional layers became a nightmare.

So, I think it's wrong to say that IBM didn't pursue things on the cloud front. There was some work on it. But the execs - not the worker bees - certainly screwed that pooch. And it'll be the execs getting the bonuses for courageously reorganizing their departments too.

Reply on news.ycombinator.com