2

I previously had negative feelings toward dedicated appliances. I enjoy getting my hands dirty and knowing the ins and outs of our infrastructure.

However, after some discussion with my colleague earlier today, I'm wondering if that's the actual general consensus.

What do you feel you gain and lose when deploying a dedicated appliance? Do you believe the time saved is worth the added expense? What are the hidden costs?

HopelessN00b
  • 53,385
  • 32
  • 133
  • 208
Belmin Fernandez
  • 10,629
  • 26
  • 84
  • 145

3 Answers3

4

Sometimes the appliance comes with specialized hardware that I'd have no possibility of duplicating, such as with the F5 BigIP hardware load-balancers. In those cases, the appliance is well worth the effort. The Cisco 6509 in our DC is a whonking big appliance with very specialized hardware, and I wouldn't dream of replacing that with software/hardware I can rip into.

Recently, appliance vendors have been offering VMWare images instead of hardware for those that just want to drop a binary blob in their ESX config instead of dealing with multiple physical servers. We have one of those right now. This makes it easier to scale up an appliance, since all you have to do is add resources it to it in the VIC.

Appliances, especially virtual appliances, are not a whole lot different than closed-source COTS. Their footprint is a lot bigger, and the debugging surfaces tend to be a LOT smaller, but not much different.

I don't mind them.

Except... each appliance is yet another discrete server that probably won't fit easily into our patching rotation and schedules. A lot of appliances really lag behind on certain OS patches, which can be a real problem. It is for this reason that I don't like using appliances unless there is no choice, or the platform for the appliance is one that isn't being actively exploited as is the case with Linux appliances.

sysadmin1138
  • 131,083
  • 18
  • 173
  • 296
1

Pros:

  • you get to ignore the man behind the curtain
  • the only way to use some specialized hardware (routers, etc)
  • Cost - in the age of integration and mass production, appliance hardware can be essentially 'bummed' down to a bare minimum BOM and then mass produced, losing an order of magnitude in price along the way. See: wifi routers

Cons:

  • configuration backups - is there a way to automate them? If not, how often do you do it? how often do you test that?
  • adaptability - if the mfr didn't envision your desired use case, you're outta luck

The best of both worlds (IMHO):

  • appliance hardware with linux under the covers + the root password
pjz
  • 10,497
  • 1
  • 31
  • 40
0

The comment on this question says it all for me:

"Our $200 Million Business crashed for 4 hours because you did WHAT???" – Stefan Lasiewski Oct 13 at 22:34

Community
  • 1
Mark Henderson
  • 68,316
  • 31
  • 175
  • 255
  • I could interpret this in two ways: Liability lies on the vendor so + for appliances. Or, you're more likely to not understand the underpinnings of your system and consequently, screw something up. – Belmin Fernandez Oct 20 '10 at 11:33
  • @Nimmy - good point actually I hadn't realised its ambiguity! I immediately felt it was #1, because chances are if you build it yourself you're far MORE likely to screw something up, and with #1 if it DOES screw up it's not your fault (probably) – Mark Henderson Oct 20 '10 at 19:28