Going VOIP

I switched our office telephone system to SIP-based VOIP a few weeks ago and I doubt we’ll ever look back. The office and data center are now connected with one touch dialing, and I just put a Cisco phone in my home on the same system. It’s all seamless. We could have data centers around the world (and I hope we do one day) all on this system.

At the office and data center, I went with Polycom 670s. Polycom touts its HD Voice technology and rightly so. They’re easily the best-sounding phones I’ve ever heard.

At home, just to try something different, I got a Cisco SPA525G. It has a nicer display than the Polycoms, but the voice quality isn’t quite as good. It’s still totally usable, though. It’s also nice how well the Polycoms and Ciscos both work together on the same SIP Virtual PBX.

I even configured one line on the customer service phone to answer our Skype calls using their new SIP feature, although I may not bother leaving that up as very few people call us on Skype and our Skype phone works just fine for that.

The great thing about a system like this for a distributed company like ours is that it makes everyone feel a bit closer to each other. I just touch one button to talk to the office or the data center, and they do the same to talk to me or each other.

Improved, easier and higher quality communications can’t help but improve any organization.

Where We Began, the history of a Dedicated Server company.

Some of you might enjoy learning how started.   As one of the leaders in the Dedicated Server industry, it’s hard now to imagine how tiny we once were, but that’s exactly how it was.  It’s a bit of an unlikely story, but I’ll be happy to relate it, anyway.

In 2005, I was in medical software when the company I worked for experienced some financial problems. I was working long hours there and was mentally exhausted. The end was almost a relief. Still, I’ve never been rich and I had to make sure Reggie stayed fed, so I spent some time considering my options. I had a couple of choices: I could try to rebuild my consulting/software-contracting business with new clients, or I could do something totally new.

The truth is, I despised software contracting. I was the product and I could only sell myself so thin. When you’re the product, you can sell your days, your evenings, your weekends and your nights, but eventually, you run out of time to sell and early in the cycle, you learn that you will never have a life.

At the same time, I had rented dedicated servers for years for my own use. I was one of the earliest customers of Yahoo Servers when they were trying that, and I’ve rented from a half dozen other companies around the world since. I even rented one of those ugly purple Cobalt Dedicated Servers from Sun back in the day. None of them were worth anything, and they all charged ridiculous prices for support I didn’t need.

I decided to try my hand at dedicated servers.

First, we hosted at another data center. I rented a cage and deployed some dedicated servers and waited for the world to beat a path to our door. Nobody knocked. I tweaked the offerings, refined things, offered new services, but nothing mattered. No takers. Then, I decided, I’m going to find the price at which people are willing to buy. Doesn’t matter what that price is. If it’s $10/mo, I want to know that. I cut prices by $10/mo until people started to buy, and I still remember like yesterday, that mark was $79/mo at that point in our history. It seemed that, at $79 for a P4 2.4 with a 100gb HD and 512mb of RAM (these were great specs for the day), people were willing to overlook that we were new and were willing to hit the buy button.

Things moved fast. My sister got involved, making it a bit of a family business, although she’s in Toronto so her involvement is remote. We quickly outgrew the cage we were renting, and moved to a bigger cage at another data center, but the next real breakthrough came when I was driving through Surrey one day. I saw this huge, beautiful and largely empty skyscraper, Central City. I wondered why such a building would be empty, and I called the number. It seems the building had been built for ICBC, our insurance company, and the newly elected government of the day had refused to let them occupy it. “Too lavish,” was the complaint, although I’ve been assured there were a lot of politics in play to nobody’s surprise. I don’t know or care.

By the time I actually got through to someone who could talk to me, the building had been leased fully to lawyers, accountants and other such companies. Simon Fraser University had purchased the first seven floors and were operating their technical college there. It was a vibrant, exciting place to be, full of bright young students and youthful energy. Still, they had no use at all for the many millions of dollars of data center equipment that were there, including huge diesel generators, a giant battery room, and multiple cross connects right on the Internet backbone.

The short version is, I told them that I couldn’t pay them millions, but I might be able to make them some pennies back on the dollars they invested. We moved in in November of 2006 and have never looked back.

We were tiny then, with just a few racks and about a hundred dedicated servers in total. Now the data center, huge though it is, is nearly full of dedicated servers in useful production and we’re looking for more space elsewhere. I’m hopeful we will have something operational by mid 2012.

Along the way, we’ve led in so many ways, including our dedicated server management system and our unique DDOS protection systems. This year should bring us some real movement in the area of KVM/IP dedicated servers. I’m hopeful we’ll reach the point soon where all servers we deploy have full-time KVM/IP units. In addition, our cloud systems are evolving and improving, and we’re looking to have a world-beating offering there within the year.

It’s been an amazing ride, but the journey has just begun.

The Amazon EC2 Cloud

Well, I’ve just spent a couple of days working with the Amazon EC2 Cloud. I have a project going on that requires multiple data centers and so I thought I’d give Amazon a try. Overall, I really didn’t have a good experience.

First, you have to choose from a confusing array of options and pricing. Most of their options prefer their own Amazon Linux which is based on Centos but is different enough that nothing worked easily for me. Even simple things like installing Webmin were troublesome (it installed ok, but none of the modules worked out of the box — every single one I used needed tweaking). You can get Centos if you want to pay for one of their more expensive packages. That would have likely gone better for me.

Second, you don’t get an IP, which is a bit ridiculous. All you get is a private IP (10.x) and you are expected to connect via a FQDN that Amazon gives you (they obviously have some host-name-based NAT going on). This works, but is a giant pain to use. You can’t even register name servers for a domain with this method. You can add IPs (up to 5 per server), but they cost $7/mo.

Thirdly, for their micro instances, you get all of 8 gb of HD space. No typo. We just switched from 500gb HDs to 2000gb HDs, and they’re giving clients 8. lol.

Finally, and this surprised the heck out of me, it was SLOW. My sites came up sluggishly and simple things jerked the screen around. Things just didn’t render quickly. I was really expecting that, with Amazon’s infrastructure, things would be quicker, but I guess they’ve sectioned things off pretty frugally.

I’m not against our competitors.  A lot of them offer excellent services, and we cooperate with many of them.  I understand that we’re not the only company in the business, and I get the appeal of the Elastic Cloud in theory. It’s a cool idea. The simple fact, though, is that it sure didn’t work for me in the real world. Our cheapest server is far faster and better, and costs less to boot.