dslreports logo
 
    All Forums Hot Topics Gallery
spc
uniqs
52

Davesnothere
Change is NOT Necessarily Progress
Premium Member
join:2009-06-15
Canada

1 edit

Davesnothere

Premium Member

Re: Where are the SERVERS ? - Summary

 
SOME MORE INFO about several providers' servers' locations :

»Re: CC Disaster

SpaethCo
Digital Plumber
MVM
join:2001-04-21
Minneapolis, MN

2 recommendations

SpaethCo to Davesnothere

MVM

to Davesnothere
said by Davesnothere:

Part of this thread's raison d'etre is to point out and clarify such info about common VoIP providers.

Still with no summary...

The part of this thread that is potentially misleading is that it tends to suggest that if your provider has geographic redundancy you're covered. That may not be the case.

- DID outages can still occur even if the provider soft-switch is not located in an outage area. Every provider listed in this thread has taken a hit to DID service on the east coast. You can move the SIP portion of call control pretty much anywhere with IP connectivity, but the gateway back to the TDM world has a physical location requirement.

- Even with geographic distribution of resources, there are still single elements of failure. Notably in the VoIP market, that's going to be the accounting database. Providers need to know if you're an authorized user, and if you have funds available to complete a call. When operating multiple nodes, that means they all have to agree on customer credential and balance information. This is typically handled by either having all call processing nodes point to a central DB, or leveraging DB replication functions to distribute the DB availability. The key problem in either case would be the introduction of data corruption.

- Depending on your provider's chosen E911 vendor, E911 services may be linked to a single point of failure.

- I've noticed at least one provider here is using GoDaddy for domain registration. If you do a cursory google search you will find it is ridiculously easy to get a domain suspended with GoDaddy. If someone were to craft a social engineering attack for a Friday night, there is a very good chance the NS records for the domain will remain pointed to "NS1 & NS2.SUSPENDED-FOR.SPAM-AND-ABUSE.COM" until they can take it up with a manager in the Abuse department on Monday morning.
MartinM
VoIP.ms
Premium Member
join:2008-07-21

MartinM

Premium Member

said by SpaethCo:

- DID outages can still occur even if the provider soft-switch is not located in an outage area. Every provider listed in this thread has taken a hit to DID service on the east coast. You can move the SIP portion of call control pretty much anywhere with IP connectivity, but the gateway back to the TDM world has a physical location requirement.

This apply to all VoIP providers, without exception.
quote:
- Even with geographic distribution of resources, there are still single elements of failure. Notably in the VoIP market, that's going to be the accounting database. Providers need to know if you're an authorized user, and if you have funds available to complete a call. When operating multiple nodes, that means they all have to agree on customer credential and balance information. This is typically handled by either having all call processing nodes point to a central DB, or leveraging DB replication functions to distribute the DB availability. The key problem in either case would be the introduction of data corruption.

That is pure speculation. Database corruption can happens as easily on a centralized setup than a distributed. At least if it's distributed, it's easy to update database resolution to one that isn't corrupted. In our case, it's distributed, and we do not have a single point of failure. Worst case scenario would be if we had to run a Read Only database while we move a master, VoIP operations would still be operational, but users wouldn't be able to modify their settings until a new master is up.
quote:
- Depending on your provider's chosen E911 vendor, E911 services may be linked to a single point of failure.
In VoIP.ms case, all VoIP servers have independent interconnection with the 911 provider, they are not dependant on a Proxy or single point of failure. Same applies for Termination and Origination.
quote:
- I've noticed at least one provider here is using GoDaddy for domain registration. If you do a cursory google search you will find it is ridiculously easy to get a domain suspended with GoDaddy. If someone were to craft a social engineering attack for a Friday night, there is a very good chance the NS records for the domain will remain pointed to "NS1 & NS2.SUSPENDED-FOR.SPAM-AND-ABUSE.COM" until they can take it up with a manager in the Abuse department on Monday morning.
VoIP uses go Daddy SSL certificates and as a partner to nic.ms, the entity that controls our domain. Again this is speculation, there are easier ways to use social Engineering and DDoS attacks to bring down a service, as we have seen recently, but I will not publish "tips" in a public forum.