MWEB Business clients' massive data loss

deweyzeph said:
Why are you using an IP address instead of a domain name to reach your server? That's just stupid. If you were using a domain name you could just change the A record in the nameserver to the new IP address and you'd have no problem.
Click to expand...

Sure, however it doesnt excuse whats happened, besides propogation isnt instantaneous genius, in which case reallocation of an ip address is quicker.
 
FatBoySlim said:
Sure, however it doesnt excuse whats happened, besides propogation isnt instantaneous genius, in which case reallocation of an ip address is quicker.
Click to expand...
You can set the ttls and expiry times of your zones and records.
Using ip addresses is stupid. Full stop.
And even if propagation takes a couple of hours, how long does it take to reconfigure every client you have?
 
Necuno said:
This can't be. Mweb always works and are never down :)
Click to expand...

MWEB Connect is not MWEB Business (IS).

And you are quite right, though we are finally starting to see some shaping as of late, which will hopefully be resolved soon. Though that has nothing to do with always working and never being down, which is true.
 
Sinbad said:
You can set the ttls and expiry times of your zones and records.
Using ip addresses is stupid. Full stop.
And even if propagation takes a couple of hours, how long does it take to reconfigure every client you have?
Click to expand...

Time to fire my dev team, you looking for a job?
 
Although incidents like these are absolutely inexcusable, they have the effect of exposing where we (as devops and developers) are cutting corners and not using best practices in the applications we build and the servers we use to host them. I'm sure a LOT of people are going to be changing their backup strategies after this mess, and people whose IP addresses have changed without warning are also going to learn some valuable lessons too. We all cut corners for whatever reasons, but sometimes an incident is a big warning about why we shouldn't.
 
deweyzeph said:
Although incidents like these are absolutely inexcusable, they have the effect of exposing where we (as devops and developers) are cutting corners and not using best practices in the applications we build and the servers we use to host them. I'm sure a LOT of people are going to be changing their backup strategies after this mess, and people whose IP addresses have changed without warning are also going to learn some valuable lessons too. We all cut corners for whatever reasons, but sometimes an incident is a big warning about why we shouldn't.
Click to expand...

Good point, at least we have backups - I pray for the poor souls who dont!
 
km2 said:
Have you read the contract from your fancy host? They probably charge you 50% more but still have it clearly spelt out that you can go get stuffed if there is data loss.
Click to expand...

Our lawyers read it, yes we have those for this sort of thing. This is pretty much how things happen in the corporate space, our clients also have lawyers reading through our SLAs to them to ensure we don't try and shaft our clients.


km2 said:
But aren't you describing a single host setup yourself? They provide the service and the backups. You're trusting them with both aspects.
Click to expand...

Nope, a single VM from our host can span 2-3 DCs in Gauteng and will be backed up on average 6 times a day between us and them.
 
ponder said:
:D

You poking fun at PostmanPot?
Click to expand...

PostmanPot said:
MWEB Connect is not MWEB Business (IS).

And you are quite right, though we are finally starting to see some shaping as of late, which will hopefully be resolved soon. Though that has nothing to do with always working and never being down, which is true.
Click to expand...

Thanks. Truth is out there :)
 
FatBoySlim said:
Time to fire my dev team, you looking for a job?
Click to expand...

Lol your dev team is pretty rubbish if they believe in IP addresses, sorry but its true.
I would never ever tie myself to an IP address unless I own the IP space in which case I can move it with my server to a new DC.
 
I don't think a lot of people realize that customers had paid MWEB Business for a backup solution and had usable backups, but when their "provisioning error" happened it deleted the customers backup's along with their VM's.
 
Murmaider said:
I don't think a lot of people realize that customers had paid MWEB Business for a backup solution and had usable backups, but when their "provisioning error" happened it deleted the customers backup's along with their VM's.
Click to expand...

Sure, but keeping backups in the same DC as your server is asking for trouble. Definitely a big no no.
 
deweyzeph said:
Sure, but keeping backups in the same DC as your server is asking for trouble. Definitely a big no no.
Click to expand...

Oh course, but unless the DC is being hit by a plane or falling into a sink hole, having backups on separate storage infrastructure the other side of the building is sufficient for the majority of people. The issue comes in when your backups are deleted along with your VM.
 
Ipwn 4 said:
Lol your dev team is pretty rubbish if they believe in IP addresses, sorry but its true.
I would never ever tie myself to an IP address unless I own the IP space in which case I can move it with my server to a new DC.
Click to expand...

Bit of a blanket statement here. Depends entirely on the situation. Many developers don't have control over their client's DNS so it can be preferable to use an IP if they are liable to update it, let it expire, etc. Obviously generally less maintenance to use a FQDN but an IP can be updated immediately without waiting for DNS to propagate when time is of the essence. Also when the application is first rolled out clients sometimes haven't worked out their domain name, etc. Calling developers rubbish if they happen to use an IP address for something without knowing the situation is uncalled for.
 
garp said:
Bit of a blanket statement here. Depends entirely on the situation. Many developers don't have control over their client's DNS so it can be preferable to use an IP if they are liable to update it, let it expire, etc. Obviously generally less maintenance to use a FQDN but an IP can be updated immediately without waiting for DNS to propagate when time is of the essence. Also when the application is first rolled out clients sometimes haven't worked out their domain name, etc. Calling developers rubbish if they happen to use an IP address for something without knowing the situation is uncalled for.
Click to expand...

There's nothing stopping developers from using their own domain name. It doesn't matter what domain is used, as long as one is used.
 
Ipwn 4 said:
Our lawyers read it, yes we have those for this sort of thing. This is pretty much how things happen in the corporate space, our clients also have lawyers reading through our SLAs to them to ensure we don't try and shaft our clients.
Click to expand...

Well maybe they need to read it again, or you need to speak to your own lawyers to understand what is actually offered by the SLA, because even the biggest cloud providers in the industry set a very small hard cap (frequently linked to the amount you've paid) for damages suffered.

And the best SLA in the world is worthless without the penalties to back it up.
 
deweyzeph said:
There's nothing stopping developers from using their own domain name. It doesn't matter what domain is used, as long as one is used.
Click to expand...

And then you end up with hundreds of A records for different clients on your own domain that you have to maintain? And what happens when you go out of business?
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X