Silicon Shecky

Exchange doesn’t like drive error, or bad blocks. Never has, never will, and while there are things that can temporarily correct the problem, new hardware is the ultimate solution.

I’ve been slowly prepping to do a migration from Exchange 2007 to Exchange 2010 at the company I work for. I’ve done my reading, come up with all sorts of bad scenarios, and basically anything else I could think of to prepare for it. Mind you, I’m not the only high end internal IT guy (Engineer, support, sales, etc…), but I’m also the only outbound tech. I had things planned out to finish the actual prep the day before a long weekend a few weeks ago, just in case I ran into any problems.

Smart thing I did that, because I ran into a major problem. I had to go to a client site, due to a printer issues. the client is a major one for the company and the directive came at 10pm in the evening from by boss to be out there the next day. This of course caused me to cancel the planned migration.

The day of the cancellation was going to be installing Exchange 2010 on the newly purchased server, and installing the latest version of Blackberry Enterprise server so that we could keep using our Blackberries. Needless to say, a few days after the cancelled migration date, our current Exchange 2007 server starts running really slow. Disk errors, bad blocks, a chkdsk cleared the errors, and I was put on the hot seat.

I explained why the migration had not happened, how I was ordered to be down at a client for a printer problem. How the amount of e-mail data will take 2-3 days to migrate, and that I wanted to do it over a long weekend. I was asked for a hard date for the migration, something soon, since slow or non-working e-mail near the end of the month was not acceptable. So I gave a date of this upcoming weekend, and went to work on getting the domain all prepped.

So here I am trying to run the Schema and AD prep on a domain where the Exchange server is in a separate site (Not domain, just physical site) from the Schema Master. Not only that but the Schema Master is a 2003 server. Yes, following Microsoft’s information of just running the Schema Prep through a 2008 server that is in the site where the schema master is located, has not worked so far.

I know I’ll get it, I dealt with this went tossing SP2 on the Exchange 2007 server, I’m just frustrated that Microsoft doesn’t even know how its own stuff works.

Everyone is talking about Cloud computing. Put things up on The Cloud. SMBs save money by moving to The Cloud. The Cloud is not everything its cracked up to be though, as one of the biggest cloud providers has recently shown.

Before we get into the heart of this op/ed type piece (I do try to use facts, but the thoughts are my own), let us take a basic look at what cloud computing is generally being marketed as. The basic idea is that you remove the server from your location, put it on the Internet through a secure host (the biggest names hosting are Microsoft, Amazon and Google), therefor giving you the ability to work from anywhere, not have to worry about server maintenance, or having an IT department( there are other aspects such as MSPs, Backup to remote data centers, and more that do not apply to this article). To quote the movie Murder by Death, “Interesting theory, one small problem. Is stupid, is most stupid theory I ever heard!”

Why is it so stupid (In my opinion). For a few reasons. First and foremost is security. Take a look at the recent problems with Google and China and you will see what I mean. The hacking, the lack of being able to harden a server yourself (or letting an IT company you can hold responsible), the lack of control. Take a look at what is going on through some of the security sites, not just the small spattering you hear from mainstream media(which will not always tell you the full story due to corporate connections). Now you might say, but I’m small why would someone want to hack me, and that is not the reason you would be hacked. It could be a side effect of being hosted on a much more visible target (again, Google, Microsoft, Amazon etc…).

Once you get past the security aspect, you run into, what happens to your data overall. Who owns it? If you got out of business, does it get destroyed properly? What if you decide to move off the cloud to a local server, does the hosting company have to keep copies of your data due to regulations? A lot of those types of issues are easily solved through contracts, but are you reading through the contract properly. From personal experience with Off-Site backups, the company I work for and our partners put in writing that the data is our clients, and if they want it destroyed due to changing services, going out of business etc… we can do that. This is just data backup though. What about when your whole server is up there?

Finally the reason the cloud is not ready for prime time is infrastructure. Mostly ISP speeds and costs. think about it, you start saving money by bringing your server up to the cloud, but find that access times to files, to e-mails, is extremely slow, and that cuts down on your productivity. The fastest you can go is going to be the slowest link in the chain.

Most businesses are still working off the T1 assumption. A T1 is 1.5Mbp downstream and upstream. This really is not a lot compared to the sizes of files, amount of data being transmitted, and other small factors such as number of people sharing that line. If you are on the cloud, you no longer have just e-mail constantly streaming in, but authentication protocols, Active Directory communications (if its a Microsoft server), Word documents, Quickbooks data (if needed) and much more. Think of it this way, the average home Internet speed is 12Mbp down, 1.5 Mbp up. Faster on the downstream, same on the upstream, which would be your clog. A T1 averages $500-$1000 per month. Home Internet costs around $30-$70 per month but does not have the Quality of service needed to be reliable for could computing. Fiber Internet is the solution (60+Mbp down and up for $1000=$1500 per month right now), but availability of it is spotty at best. Until this bottleneck is fixed, no matter how secure it might be, or guarantees about the data ownership are resolved, I cannot recommend could computing.

The biggest thing to realize is that there is give and take in everything. To really come up with savings, you have to figure in items such as security, lost time due to connectivity, plus you still need someone to be able to take care of your local PCs. A good local IT consultant in the long run is still a better option for most SMBs. A Managed Service Plan with a local IT firm is probably the best, cause its a one low cost solution that covers most everything, and you can budget for because the cost is locked in for the length of the contract. Think about that before you go cloud hopping.

So after all is said and done there is still more to do.

You all should find some sort of Backup and Disaster Recovery plan for your clients. I’ve started to look at the one we offer at work and realize that it can be a life saver for any business.

Think about it, a good one will make sure that you are protected and can be back up and running in an emergency within 48 hours. Have only up to 48 hours of downtime is short as it is. Less business lost, less downtime, quicker recovery for the business. It should be a no brainer, which is why I’m surprised at how many businesses decide its not worth the money.

I guess we can’t win them all, even when we try to be proactive.