Server Hosting Options - A Rough Guide

Potential Customer: “I have a startup/website/application in development and launching soon and am looking into hosting options. How cheap is hosting with UnitedLayer?”

As someone who frequently hangs around and associates with entrepreneurs and small growing businesses, I get this question nearly 2-3 times a week. In truth, it is a very good question as there is a lot of confusion around “uptime guarantees”, “Cloud Computing” and exactly what someone gets when they have a “managed server.” This is mostly due to the ridiculously large marketing dollars companies like RackSpace, MediaTemple, GoGrid (aka ServePath), Amazon Web Services and others have dumped into the Web 2.0 space and Blogosphere coining catchy and sexy phrases that make it sound like all your hosting troubles will be solved in the glorious, wonderful, happy, and lucid Cloud.

Below is my attempt at answering these questions as objectively as possible having spoken with hundreds of developers and systems engineers who have been through the ringer leveraging almost every conceivable method of delivering 1’s and 0’s across the internet. Though I work for a Hosting and Colocation provider myself, I fully understand and can admit that UnitedLayer is most definitely not the answer to everyone’s needs. …and for what it’s worth, I am also not subject to the bias I have seen in most developers and systems administrators who have spent their whole life doing one thing, and therefore believe it is the only and best way for everyone to do things.

“I HAVE LITTLE EXPERIENCE SCALING APPLICATIONS AND HAVE NOT SPENT MUCH TIME LEARNING SYSTEMS ADMINISTRATION”

If this describes you, or the person on your team in charge of managing your application and the hardware it exists on, it would be extremely wise for you to learn about Amazon Web Services.

Leveraging AWS, you’ll be able to:

1. Only pay for what you use (CPU, RAM, HD, Throughput)
2. Not be limited by a physical hardware components and have to upgrade your server if/when traffic starts scaling
3. Start generating hard metrics on utilization of CPU, RAM, HD, Throughput/Bandwidth so that you can accurately predict and forecast hardware and transit needs as your user base and traffic increases

Once you’ve got a firm grip on how each new 1,000 or 100,000 new users/transactions effect your site, you have a few options depending on how important Security and Stability are to your production environment.

• Option 1: Stay with AWS
• Option 2: Migrate your production environment to a manged or dedicated hosting provider
• Option 3: Buy a bunch of hardware and colocate and manage your production environment yourself

If security and stability are low on your scale of priorities, go with Option 1. If security and stability are important to your production environment however, option 2 and 3 are really your best options.

If security, stability AND cost are important, Option 2 is really your only option.

Why is Security and Stability an Issue for Cloud Computing?

For an excellent conversation on why Security is a HUGE issue within the Cloud Computing platforms currently being offered, you should definitely READ THIS and talk with or read comments from some other senior IT executives at Fortune 1000 firms who have to walk through an IT auditing process.

In regards to Stability, Cloud services like AWS are definitly getting better, and starting to wrap Service Level Agreements around their offering so that if they do bite it, you get something (though always marginal) in return. It is important to realize however that Cloud Computing is still a relatively new science with few generally accepted standards, limited experience from those actually behind the scenes architecting how to physically scale this to accommodate millions of users, and a BOAT LOAD of marketing dollars behind it. AWS went down just last week in fact, AGAIN, and admits they don’t have everything perfected yet. (Personally I don’t think anyone will “prefect” it any time soon or even think it even can be “perfected” on the type of scale it’s on.)

What is the Difference Between a Dedicated and Managed Server?

Simply put, a Dedicated Server is a server that YOU DO NOT OWN but effectively rent on a monthly basis that is pre-configured to your desired specifications (example:  Quad Core, 2 GB RAM, 2 x 250 GB SATA Raid 1, Debian OS, Apache) and then full root access to the machine is turned over to you.

For these types of offerings, hosting providers try to lock you into a 12 month contract as they likely won’t see a lick of profit until @ the 8th to 10th month you’re actually using the machine.  And, for what it’s worth, you would rather work with a company that is in fact MAKING MONEY, as this means that they have the capital to invest in building out their hosted environment leveraging best of breed business practices that in the end will only ensure your machine never looses power or a connection to the internet.

A Managed Server is also a server that YOU DO NOT OWN but effectively rent on a monthly basis that is pre-configured to your desired specifications.  Access to your Operating System and “full root access” however is managed by your provider.  For some visual representations of what this actually means, check out HERE and HERE.

When you have a managed server, you are effectively paying a little extra to have an outsourced person or team serve as your systems administrator for your server(s).  They are also, in theory, suppose to help you architect your environment as your needs grow over time.  For example, you may eventually need to add a Firewall or Loadbalancer into the mix or separate your DataBase from your Web Server as traffic grows, or add a SAN or separate storage device for backups… and your systems engineer is suppose to PROACTIVELY manage your environment in this way.

Again, same theory applies, in that you want to work with a company that is in fact MAKING MONEY, as this means that they have the capital to invest in not only building out their hosted environment leveraging best of breed business practices that in the end will only ensure your machine never looses power or a connection to the internet BUT also hire systems engineers who are both good at what they do and PERSONABLE!!!!

How many engineers do you know that are brilliant at what they do, but totally incapable of explaining anything to you without either you or them losing their sh*t?  Why does this matter?  Try calling in at 3am and working with someone over the phone that you likely have never met before to get your site back up and running.  (Personally, in all the years I’ve been working with and around engineers, I know very few, and have made a point to befriend those that have both a right and left functioning brain)

“OK, SO I GET IT, AND NEED A HOSTING PROVIDER FOR MY PRODUCTION ENVIRONMENT… NOW WHAT?”

If all you want/need is a few dedicated servers, there are 100’s, if not 1000’s that can offer you this service.  Of these however, there are only a hand full who actually know what they’re doing on the backend and have architected an environment that you can actually grow into without causing major headaches, downtime, ridiculous expenses and/or a jerryrigged system.

What you should be looking for is the following:

1. Blended and Managed Transit - Provides redundancy in your network so that if one carrier goes down, traffic simply routes across the next best path from another provider
2. 24/7 Systems Support - Best way to know if the company your planning on hosting with is playing in the big leagues is if you can call in at 3am and speak with someone who actually knows what the status of your account is, if you’re a registered person on your company’s contact list, and can actually help resolve your issue within a timely manner while keeping a smile on their face and not sounding like they just woke up.
3. A Stellar Uptime Record - Great thing here is that if a provider goes down regularly, you can count on other geeks harping about it online, so definitely check references and check forums online for what other people are saying about a company’s uptime record.
4. A Clean Facility - If there’s any way you can actually see the facility you servers will be hosted within, do it.  Hosting is extremely technical and complicated stuff when done correctly and though a company might have a brilliant looking website, their datacenter might be a POS, meaning they aren’t interested in investing in maintaining standards, their equipment, or truly SERVING their customers once they’re signed up.
5. Tools That Make Your Job Easier -  Documentation and accessibility are key, as a company that invests in the tools that allow them to communicate effectively with their customers and their customers to communicate effectively with them is also a company that, more times than not, communicates effectively internally and has their operations and suppor team running a happy and motivated ship.
6. Experience - Check the BIOS of the key executives, especially those running and managing the onsite support teams and systems teams.  If they’ve been there and done that successfully before, they’ll likely be able to repeat the process.

“OK, SO WHO DO YOU RECOMMEND?”

Well, since you asked ;)

If I only needed a dedicated single server to host some basic websites with decent traffic, I would work with SoftLayer.  These guys successfully grew ThePlanet from scratch and bounced when the company was bought out and new management came in.  They’ve now built, over only the past few years, a rock solid infratructure that is solely dedicated to servicing the specific nitch of those needed DEDICATED SERVERS.  They DO NOT do Managed Servers however, or any other managed infrastructure, so if you need someone to manage a firewall, loadbalancer, switch, router, ect. you’ll want to look elsewhere.

If you need someone who can not only provide you with a hosted environment, but also provide a level of systems expertise so that you really only need to worry about your application layer, I would highly recommend giving UnitedLayer, RackSpace, ThePlanet and Savvis a call. There are likely other providers worthy of being on this list but I can only speak from my experience.

…and here’s my sales pitch :) …

At the end of the day, I truly believe UnitedLayer has a pretty stellar team of extremely technical and forward thinking people that are actively trying to make the service and services we offer best in breed.  This being said, we don’t have a staff of 1,000’s like RackSpace and Savvis or a marketing budget of $10-100,000 a month like most of our competitors because we are still a privately held company, having closed @ $7M in revenue last year and received NO VENTURE CAPITAL.

This means we’re still working on getting our “3am rule” service up to something to be proud about and still developing many of the communications tools (rolling out in the next few months) allowing our customers to find any information they need and do pretty much anything they want remotely.

On the flip side, we don’t have 20,000+ customers and our people know all our customers by name.  Our hosting environment has also had only one minor network outage in the past 2 years that was very short lived.  We also know the specific types of companies we want to be working with, given most of the people at the company have been entrepreneurs at one point in their life and can spot talent and great ideas when they see it.  Really, this means we prefer to work with fewer companies that have great ideas and growing needs, than a ton of customers who have minimal needs and are just starting to figure out a business model.

Well, if you’ve made it this far, I hope this data dump helped clarify a few things.  If you have any further questions or comments, feel free to contact me directly!

Best,

Sean