Monthly Archives: May 2009

Is it AT&T, Parma, or Bad Luck?

The company I work for moved their corporate offices (with which I am affiliated) a little more than a year ago from Cleveland, Ohio to just down the road in Parma, Ohio.  After the move we switched from a single, dedicated T-1 circuit to a series of MPLS circuits that utilize managed (i.e. AT&T owned) Cisco routers.  At our two other MPLS sites, their circuits have been up constantly for the past year while the corporate offices in Parma are now experiencing their third major outage.  Since many major IT services originate from our corporate office this is an inconvenience to say the least.

I was always cool to MPLS since I never cared for the throughput of it’s precursor, frame relay, however AT&T made a strong case which included nice technical capabilities (class of service mostly) and high cost savings at least in the short term.   I’d guess that I’d be mostly satisfied with the service were not for it’s horrible reliability which negates whatever benefit the product might have.  I don’t think my expectations are too high; our Internet circuit, which is still in Cleveland, has only had one major outage in the last ten years (though to be fair that’s over the course of four different circuits, but still).

Insult on injury is the fact that AT&T bought Ameritech some years back and the tech support line for the MPLS circuit will constantly refer to the call being referred to the ‘local telco’.  THE LOCAL TELCO IS YOU AT&T!  I’m amazed that after all this time they haven’t thought of some sort of intra-company integration that might get me an update on my call sometime before thirty minutes after the problem has been fixed.  It’s a shame because I remember the Ameritech days when I would actually get to talk to the person who was fixing my issue rather than someone in a far off land staring at a terminal.

Every time this happens I know it’s not my fault, but I was the one who sold it to my company.  MPLS in Parma?  Thumbs down.  Alternatives?  None really, especially if you need class of service on the line.

Update 5/29/2009: I want to tip my hat to our current rep who did us right on pushing for and obtaining a full credit that will help cover the costs of this outage.  This will definitely make it easier to pitch AT&T products to management in the future, thanks!

Upgrading Customized Exchange

So if I have created custom e-mail address policies and/or custom address lists the upgrade process will not upgrade them?  It’s stuff like this that makes me think that Microsoft still sells 90% of their product to five person shops that never change or monitor anything and the other 10% to gargantuan enterprises that can afford to hack around the missing pieces (in this case by crafting a complicated custom script).  Equally, it’s unfortunate that and the needless interface changes to both the Exchange and IIS management consoles detract a little bit from some of the nice new enhancements to those packages (in particular, I’ve already had users tell me that they liked the ability schedule out-of-office notices with Exchange 2007).

And no, I don’t think that the scripting back-end is what goofed up the interface.  IBM’s SMIT did the same thing but it was easier to use (ASCII menus and all).  It’s the inevitable learning curve with a new package I guess; I’ll just have to be patient!

ArcMail Defender

I’m torn on this device. 

When we were first looking into an e-mail archiving solution I thought about ‘brewing my own’ since I didn’t figure that it could be that hard to use the Exchange journaling option to dump e-mail out to a Linux box running MySQL (which as near as I can tell is what the ArcMail Defender box is).  However I would be running the risk of having it go the same road as all my other Linux projects: a stable, low maintenance solution that’s preceded by a horrible, painful amount of configuration time (a time so painful that I might not finish it).  In this case that was time I didn’t have so we elected to go with the most cost effective solution, the mini-ATX ArcMail Defender.

I’ve have had to use the software that comes with the device twice to retrieve e-mails and it proved a trouble free task.  The box also helps me sleep easier at night since I know that there’s yet another e-mail backup running, one that runs all the time.  The only issue I have is the sketchy support.

If you go to the company’s front page you’ll notice (at least at this time) that there’s no link for support.  No support forums (usually the only source for decent online support anywhere anymore), no e-mail links (it’s support@arcmail.com) and no phone number (it’s 1-888-790-9252) all of which give the impression of a fly-by-night operation.  On top of that I’ve had four hardware issues with the box, three of which required ordering parts. 

I should note that the first issue occurred when we initially received the device.  While checking out the box to see what it was made of (it’s made of off-the-shelf parts) I noticed that the RAID card was improperly wired and seated.

I guess it’s like a lot of tech items: a dream when it works, not so much when it doesn’t.  I think I’d feel a lot sunnier about if ArcMail would turn their game up a notch.  Generic parts are fine, so long as they don’t fail and less rigorous tech support is acceptable, so long as your customers can get to it.

*What instigated this post was a hard drive replacement that took more than two weeks to get to me (they blamed UPS, which might be legit I guess) and subsequent late call back to tell me how to rebuild my RAID 1 set using the opaque RAID tools that come with the system.

Why?

I applied an update to a virtual server sitting on Microsoft’s Hyper-V and rebooted it, but it didn’t come back up.  When I tried to start the virtual machine I received the error “Microsoft Emulated IDE Controller (Instance ID {GUID}): Failed to power on with Error ‘General access denied error'”.  The fix was to remove the hard drive from the virtual machine and re-add it.  What is up with that?!?

Backing Up Hyper-V

Microsoft’s Data Protection Manager (DPM) is a dream when it works, but I spend way too much time wrestling with it to make it work and the reports it issues to track jobs are worthless.  The whole experience reminds of Arcserve circa 1997: poorly made technology trying to do too much. 

It seems as if the software doesn’t miss a chance get a protection status of ‘Recovery point creation failed’ or ‘Replica is inconsistent’.  Why did it ‘fail’?  Why is it inconsistent?  Disk space seems to have something to do with it most of the time, but other times it just seems as if the software just got grumpy.  One of our servers is in an even worse state since any Windows Server 2008 backup (either the built-in or DPM) ends with a not entirely unknown  0x0000007E blue screen stop error. 

Is DPM going to be like Arcserve where it took CA six years to write a product that wouldn’t crash every time I went to back up Exchange?  I hope not.