2 ms·
I find it kind of odd, but IBM pretty much wrote the book on building reliable data center services. I've met the CEO/GM of SoftLayer (Lance Crosby) and he is a
by ChuckMcM 12y ago
I find it kind of odd, but IBM pretty much wrote the book on building reliable data center services. I've met the CEO/GM of SoftLayer (Lance Crosby) and he is a smart guy so I'm guessing he will have a lot to add to the bottom line going forward. But it did not seem like Lance was working with IBM so much as he was working for IBM, and that strikes me as something that will be corrected once the integration is better.
VPS hosting is a commodity business, and Digital Ocean is driving every penny out of it (they are sort of the Dell of VPS it seems). But the "cloud" is a services and consulting play I believe. A couple of the things I learned at Google was that uniform server design didn't work well, but integrated compute and services did. A Gmail cluster was a thing to behold. The challenge is that if you're building petabyte and exabyte scale infrastructure you can't have your "storage cluster" over here, and your "compute cluster" over there, ever move a petabyte of data over a 10GB link? It takes forever. So your goal has to be to build an API/Services rich environment in a high bandwidth fabric but to ruthlessly use locality as a "feature" in your scheduling.
With Softlayer's gear, and IBMs resources that is really doable. Remember that IBM is still going to have 12 - 13 billion in free cash flow. So they have the horses to build infrastructure just as fast as Google does (which calls out that expense in their quarterly reports), but IBM also needs the services infrastructure play which is something they are really really good at. I'm not counting them out until the decide to divest themselves of SoftLayer :-)
- MichaelGG 12y agoWhat do you mean by uniform server design not working out? You mean without locality planning?
- ChuckMcM 12y agoThe big challenge is storage. If you don't treat storage as a first class service and instead make it part of the standard 'atom' your storage requirements grow faster than your compute requirements. And while you get some relief by replacing the storage component with denser ones, you always seem to end up with high priced and under employed compute capabilities going to waste while it provides access to data on some spinning rust. I was an advocate of variation in the form of 'storage focused' and 'compute focused' atoms but at the time the mantra was 'identical blocks.' Never got much traction on that. The trick is of course mixing nodes the the mesh of services, that aspect of it really benefits from a colloidal solution of resource nodes :-)
- nknighthb 12y ago> But it did not seem like Lance was working with IBM so much as he was working for IBM Ugh. I was a TP and then SL customer for years. Having somebody controlling Lance strikes me as a very bad idea. Maybe a good way to trigger the founding of another Dallas-based provider. I kind of wondered about this when IBM bought them, but I'd been entirely Linode-based for a while so didn't have much stake in things.