Comment by PaulHoule | Hacker News Reader

Comment by PaulHoule | Hacker News Reader

Parent Full thread

It hasn't bothered me.

I quit Softlayer because every time I sent a change order in something awful would happen. At one point the issue tracking system broke in a way that I could not put in new tickets and got to talk to four different people until I talked to a wizard who punched a few commands into the SQL monitor and told me he saw something "amusing".

I had a "near miss" at data loss because one of their techs botched adding another hard drive to the machine, plus I was dealing with an expensive and balky backup system so I immediately moved my data into S3 then all the servers into EC2.

Softlayer had a crack sales guy call me to try to get me back and I told him I had a day job and a night job and I don't have time to talk to minions to fix the problems they make for me. He brought up the egress cost issue and I told him flatly that "I make $1000 a month in ad revenue and I pay $30 in egress charges so I don't care."

My business situation has changed in many ways since then but I'd say that my egress charges tend to run between 5-10% of my total spend so it is not a concern for me. If there is any AWS service that I don't like the value of it is RDS and I am mostly off it since I have been using SSD-backed instances and first running local copies of MySQL and then ditched MySQL.

Softlayer is the product of multiple provider rollups and it shows. Some of their techs are solid, but others not so much. Their processes can be pretty whacked too.

One rollup just prior to Softlayer was ThePlanet and their efforts to onramp us from their vanilla dedicated to their new managed service several years ago was a complete debacle. And our environment was super simple--only 5 boxes on a rack and a couple network devices.

But, the guy who was "leading" the effort was completely inexperienced. When we finally announced that we'd had enough, they brought in more management and senior tech guys to advise and save the deal (they were angling for an investment so wanted to book more customers before quarter's end; thus our little business mattered).

They talked us into staying and one of our conditions was that they replace the "lead". Oddly, they asked if they could leave him in place because he was "a young guy, just getting started, and the blow would set him back". Of course, I felt for the guy, but that struck me as a horrible thing to ask of a customer. Didn't want to hurt him but had to insist nonetheless.

In any case, we stayed with them for some time at close to legacy prices on fairly dated metal, primarily because we didn't have time to switch. Over that time, services became decidedly "less managed", especially after the rollup to Softlayer. Of course, by then they were also pushing their cloud. When we finally found time to switch, we moved to AWS and never looked back (except in relief).

We cut our costs by two-thirds. Better, it struck me that AWS's automated processes are an order of magnitude better than the "managed" services we were by then receiving from Softlayer.

Can I ask when you moved off of soft layer? I ask because my employer is considering them as an option and this isn't the first time I've heard softlayer horror stories.

Not the OP but we have had a hell of a time with SoftLayer and have been doing everything we can to move. If you have a large ops team (or a team that isn't busy) and you go with a well planned physical server build-out (make sure you get them in different pods) you _might_ be ok.

We have been working on a POC in AWS and it's such a breath of fresh air. Things work as advertised, the provisioning process is quick and everything can easily be done with an API call. No more waiting for support tickets. The freedom you get with your network routing in an AWS VPC is worth it to me.

At the risk of sounding too much like an AWS fanboy we are actually looking at scrapping it all and just bringing everything in-house running on our own physical boxes with some type of hypervisor on top.

> At the risk of sounding too much like an AWS fanboy we are actually looking at scrapping it all and just bringing everything in-house running on our own physical boxes with some type of hypervisor on top.

We have done just that, with Opennebula (qemu,kvm; networking built on top of openvswitch and storage from distributed iSCSI NAS appliances). It's a perfect middle ground. All the easy provisioning niceness, less than half the price of AWS.

Reply on news.ycombinator.com