Programming Reality

Life in C#
posts - 81, comments - 14, trackbacks - 348

My Links

News

Article Categories

Archives

Post Categories

Image Galleries

Blogs

CRM

Email Down: Kicked and beaten within an inch of it's life

Yeah I'm slightly pissed. Our email has been down still after the 24 hour mark. I understand why it takes 24 hours to propagate the DNS changes. DNS servers cache responses so you normally have to wait until the cache expires before it tries to query the name server again for the record you're looking for.

In this case the MX record is screwed again pointing to the wrong place again. All changes will take yet another 24 hours to propagate through the internet. The problem is I don't know who handles the DNS changes so it could be done now, or whenever our web developer gets back from wherever the hell he's at. He said he took his computer but I guess I'll believe that if/when he makes the change.

This makes me love the internet and the standard known as DNS caching. I understand it's practical use but if it takes 24 hours for changes to get where they need to be then something is wrong. In an economy where minutes are precious, waiting an entire day for something isn't practical any more. This may have been a necessity when name servers were all ran on fractional 128k T1 lines but that isn't the case now.

Turning off DNS caching can be problematic though. I don't think it should be turned off I just think that servers should cache records for far shorter periods. This eliminates a DoS attack from being performed through querying DNS records but keeps changes from taking a full 24 hours. You could have the same results caching records for 3 hours that you could if you bumped it up to 24. DNS queries are incredibly fast and have very little payload associated with them. A standard query is only a couple of bytes of data so pulling the information from the internet a little sooner won't shut down your ISP.

 

Honestly a lot of the standards of the internet are in need of a revamp. DNS is good but it needs something a little more robust. Email is wonderful but due to the nature of it, spam and malicious intent is severely easy to pull off. Email “Caller ID” isn't going to be the answer I think. Revamping the SMTP protocol is most likely going to be the only real way we're going to see relief from problems that plague email. That becomes a huge issue though as RFCs aren't easy to define or get implemented into everyday usage. Even if an RFC of said changes were done today it would still take years for the world at large to incorporate it into their products.

There is no quick fix for my problem or a way of keeping it from taking 24 hours every time a problem happens. That may be acceptable back in the Stone Age but we're in a time and place where transactions happen within a fraction of a second. If I have to wait 24 hours then I'm severely behind everyone else.

Print | posted on Wednesday, August 11, 2004 1:48 PM | Filed Under [ Information Technology ]

Feedback

Gravatar

# re: Email Down: Kicked and beaten within an inch of it's life

Internet naming isn't exactly something that changes a lot. There would be zero benefit to decreasing the cache time.

If your work is that bloody critical, there's always the phone and IM. Besides, email doesn't guarantee when or if the email will even make it through. If you're relying on it for critically time sensitive transactions, you've got bigger problems...

<sheesh>
8/12/2004 9:26 PM | David Totzke
Gravatar

# re: Email Down: Kicked and beaten within an inch of it's life

We only have internal IM which we use to communicate other than phone. It's useful when say someone is talking to a client and can't be reached.

We don't do outside IM. There's a host of reasons for this. One major one being customers would most likely feel they could reach us 24/7. Also it leaves a distraction for our employees (like playing spider solitaire isn't?).

You can't send a Word document or PDF file over the phone. You could use snail mail for this but waiting a week for material is far worse than waiting 24 hours. I suppose things could be worse in that regard.

I never disconnect anything computer related to a non computer related thing. I would never ask someone to Fax me a Word document. I don't have the time, energy, or patience to scan said document back into Word. It would work but faxing also leaves those lovely little tag lines and crap a normal Word document wouldn't leave (I know, picky, but it still proves a point).

Your arguments are valid and I do agree somewhat. DNS changes don't occur often but when they do occur wouldn't you rather wait 3-6 hours as opposed to 24? A DNS query isn't that huge of a packet size though so caching them for longer periods of time doesn't make that much sense either. The only time decreasing the cache time makes sense is when you run into this same problem.

Here's a simple solution though I don't know if it'll be practical: Have a setting on my DNS server that says I've updated my information. When outside servers get a DNS query on my domain they should query my DNS to see if the update flag has been set. If it has been set, drop the cache and rebuild it right there. If an update hasn't occured, keep your cache and operate normally.

This may be just as bad as having no cache at all though because it is performing a DNS query. It will return a small result though and probably doesn't need near the packet size of a normal query.

You don't think of this as a problem until you run into it twice in a row. That's the only reason I'm even saying anything or pissed about the whole ordeal. If this was fixed Monday I may think nothing of the 24 hour thing though it would have still stuck in the back of my mind and pissed me off eventually.
8/13/2004 12:42 PM | Jeremy Brayton
Post A Comment
Title:
Name:
Email:
Website:
Comment:
Verification:
 
 

Powered by: