dslreports logo
 
    All Forums Hot Topics Gallery
spc
Search similar:


uniqs
2639

Blunderbuss
@fastwebserver.de

Blunderbuss to PX Eliezer704

Anon

to PX Eliezer704

Re: [Voip.ms] Survey / Feedback on restructuring Geo Locations.

Your premise is clearly false because they have servers in TEN (10) different US and Canadian cities. That's way beyond what's needed for reasonable redundancy.

It isn't just for redundancy; as Martin has said, it is easier to deal with a problem when it affects only 10 percent of the users rather than 100 percent.

[Now, I'm sure that someone will jump in and say that CallCentric has the opposite geosituation....But they do need to continue to work to address this.]

Why would they bother when according to them, or maybe it is according to posters here, it is only a storm occurring once in 100 or 200 years that will ever cause them a problem?
PX Eliezer704
Premium Member
join:2008-08-09
Hutt River

PX Eliezer704

Premium Member

said by Blunderbuss :

It isn't just for redundancy; as Martin has said, it is easier to deal with a problem when it affects only 10 percent of the users rather than 100 percent.

Why should servers be going down at all?

When did you hear of servers going down at Voipo, CallWithUs, Localphone or Voxbeam, Ooma, Anveo, Junction Networks, Packet8, and so forth? Not with any regularity, for sure. Maybe once or twice at most in several years.

The funny thing is that another company that had a lot of problems with server issues was [ViaTalk]---and THEY believed in having their servers widely scattered too.
MartinM
VoIP.ms
Premium Member
join:2008-07-21

1 edit

MartinM

Premium Member

Let's not get this off topic and derail the thread from its primary objective. All providers have problems at some point. (Hurricane, Data Center failures, [D]DoS).

We're seeking feedback from active customers about their opinion of the possibility of having fewer POP's, not another debate about geo location vs call centric, this has been beaten to death by fanatics of both sides.
Mango
Use DMZ and you get a kick in the dick.
Premium Member
join:2008-12-25
www.toao.net

1 edit

Mango to OZO

Premium Member

to OZO
said by OZO:

directing RTP traffic using shortest distance (in Internet terms) between SIP client and RTP host, serving corresponding DID. If it's not, please do it ASAP. It will improve quality of voice for us, customers. Thanks.

Though I agree with you, I suspect they will not do this and I understand why. Making this work properly requires significantly more difficult configuration on the user's side. They likely believe (correctly) that this would increase support load significantly, and also bad PR when people blame one-way audio issues on them, when they're actually caused by a misconfigured ATA.
amd64now
join:2007-04-01
canada

amd64now to MartinM

Member

to MartinM
I think this is a great idea. Automatic load balancing and failover would only improve the service.

As a customer, I sometimes wonder when I have issues if it's server related. For example, the server has too many customers connected. Load balancing would solve this issue.

Like some customers, I have experienced loss of service due to a pop issues. This could have been a result of a server issue. A server fail over would solve this issue.

MartinM, if this plan is accepted, what would be a rough time line to have the geo locations established?

Blunderbuss
@fastwebserver.de

Blunderbuss to MartinM

Anon

to MartinM
My feedback is that I would like to see two or three servers that will automatically failover and work in such a way that I do not have to change the settings in my ata or on the voip.ms panel when the connection changes to another server.

Each server will have a potentially infinite capacity simply by plugging in another drive or box or whatever it is that it takes to expand the capacity.

If you get a ddos attack your super servers will hardly notice it; your customers will not notice it at all.

The servers will be as reliable as a pots line.

I will be happy to pay higher fees for this.
DSLR98004
join:2012-07-16
Bellevue, WA

DSLR98004 to MartinM

Member

to MartinM
MartinM: I support your plan. I live just outside of Seattle, but Houston is my primary POP (with Los Angeles as my fallback).

gerick
join:2001-01-17
San Antonio, TX
Google Wifi
Obihai OBi200

gerick to gweidenh

Member

to gweidenh
said by gweidenh:

Go with Dallas for your Texas location.

Most Texas ISP traffic is routed through Dallas before it hits the wild wonderful internet.

Ditto this. Anytime I check speeds with AT&T, it is always shorter ping time from San Antonio or Austin to Dallas than it is to Houston. I think that there are larger pipes out of Dallas.
lilarry
Premium Member
join:2010-04-06

lilarry to NefCanuck

Premium Member

to NefCanuck
said by NefCanuck:

One other thing I would strongly urge you to consider if a consolidation of POP's is in the cards. Automatic fallover in case a particular POP goes down and one central place for all voicemail for the POP's.

Dittos to this. Even now, when we "server-hop", or when you reroute when a server is down, the voice mail issue becomes a problem. However you do it, the same voice mail accounts need to be accessible from all servers. The same is true for Internal extensions - they should be able to work across servers. And automatic failover would also provide some additional peace of mind.

mau108
Mau
Premium Member
join:2001-10-07
Thornhill, ON

mau108 to MartinM

Premium Member

to MartinM
yes please setup one location for voicemail! switching servers and having a different voice mail greeting is annoying as hell!
phoneuser
join:2012-12-19
New York, NY

phoneuser to MartinM

Member

to MartinM
Thanks for the opportunity to comment on your plans. Here are some random thoughts to add to the mix from a long-time customer and forum lurker.

Consider having exactly two carefully chosen POPs. Expose them with separate names in DNS at first for simplicity, but prepare for the time when you can move to one name and control their use via DNS.

* More POPs means more to manage and keep in sync. One is "easy", two is usually "hard", but anything more than two is usually "much harder". Two POPs (each provisioned to handle your entire load) gives you site redundancy, is simpler to manage, and, importantly, is easier to get exactly right.

* Don't sweat the geographic diversity issue too much. Much more important is basic datacenter reliability and quality/diversity of connectivity. And don't rule out Las Vegas as a candidate, with its low earthquake and weather vulnerability, plentiful power, and great connectivity.

* SRV records are great in principle, but we've all learned recently that real-world client support for them is spotty at best, and incorrect at worst. Good old-fashioned round-robin A records with shortish TTLs are almost as good in principle, and often better in real-world practice.

* If you do use SRV records (as well as A records), don't make the mistake of returning more than will fit in an untruncated UDP response, even though doing so is perfectly "correct". Too many client resolvers (incorrectly) don't retry over TCP, and too many client firewalls (incorrectly) pass DNS requests only via UDP. If you absolutely need lots of hostnames, then define multiple A records per hostname, or use shorter hostnames (since SRV record hostnames "MUST NOT" be compressed), or something.

* Set as a longer-term goal making POP choice functionally irrelevant by syncing voicemail between the POPs, decoupling DID routing from user agent registration (like Anveo's Geo POPs, I think, but chosen and controlled by you), and much more that I'm sure I'm missing.

* Eventually, go further still by using a single DNS proxy name, and respond with the SRV or A records you want to based on load, server status, geographic estimate of client location, or whatever else you choose. Take this out of the hands of your customers. Some will complain that they want to choose their POP based on the quality of their routes to and from a particular POP. Some of this can be mitigated by your datacenter choices and the diversity and quality of their connectivity. It's not a perfect solution, but you need to control your service, and you need to eliminate your customers' need to fiddle with their endpoint and fiddle with your user portal to perform POP failover. I know you already do some of this when you want to direct customers away from a troubled POP, but you need to solve the resulting DID routing and voicemail issues, and you can go further and really control your network yourself, and do so better than most of your customers can.
OZO
Premium Member
join:2003-01-17

OZO to Mango

Premium Member

to Mango
said by Mango:

said by OZO:

directing RTP traffic using shortest distance (in Internet terms) between SIP client and RTP host, serving corresponding DID. If it's not, please do it ASAP. It will improve quality of voice for us, customers. Thanks.

Though I agree with you, I suspect they will not do this and I understand why. Making this work properly requires significantly more difficult configuration on the user's side. They likely believe (correctly) that this would increase support load significantly, and also bad PR when people blame one-way audio issues on them, when they're actually caused by a misconfigured ATA.

Latency becomes very important issue, especially when number of calls to/from cell phones is increasing. From my experience, quality of the voice could degrade dramatically when cell phones add their big portion to the overall delay. That's why routing of RTP requires special attention and careful optimization. And that's why I asked VoIP.ms to offer direct media (or bypass media) mode option for those clients, who can handle it (have equipment and needed configuration). In other words, I don't mind to connect to a SIP server, located even on the opposite part of the Earth. But I do care very much about the latency, added to the voice channel (RTP traffic).
chrisedwards
join:2008-05-17
Calgary, AB

chrisedwards to MartinM

Member

to MartinM
This is still a fairly recent thread so I thought I'd add my thoughts:

1. I don't see a Europe POP mentioned. We have employees travel frequently to Europe and it's nice to have a European POP that they can reliably connect to while overseas. I haven't tried the New York POP though, so it's possible that in various locations in Europe the NY POP would have sufficient pings. In any case I'd love to see a European POP kept.

2. I like the New York POP selection. New York has good connectivity to South America. I get 100-120ms pings to Sao Paulo in Brazil. While higher than what'd you'd see domestically, it's workable.

3. I like Los Angeles POP which has good connectivity to Mexico (~60ms to Puerto Vallarta).
papaskitch
join:2007-02-10
Woodbridge, ON

papaskitch to MartinM

Member

to MartinM
I'm generally a very happy voip.ms customer, except when it's broken

If the "mega-pops" improve reliability, I'm in.
Jonathan987
join:2006-03-17
North York, ON

Jonathan987 to MartinM

Member

to MartinM
Hi, as long as there is a POP in Canada for the Toronto/Montreal customers it would be good. But there needs to be more capacity given to your Montreal servers if you plan on making that the main connection for your Toronto customers.

The reason I don't mind having to connect to Montreal servers is that the ping is just as low it was to the Toronto servers because iWeb (I'm assuming your Montreal hosters) has full unconditional peering on torix.

Hope this helps with the decision!
MartinM
VoIP.ms
Premium Member
join:2008-07-21

MartinM

Premium Member

This idea has been put on shelves for the moment and reducing the number of locations is no longer part of our short term plans. We want to thank everyone who provided feedback in this thread and via private messaging.

We're busy updating Montreal, Toronto, Chicago and NY with a new infrastructure that will allow for transparent scaling in the future.

We're also looking to deploy a POP in Vancouver for west coast based Canadians.

Regards,
dirt_diver
join:2012-12-21
Louisville, KY

dirt_diver to MartinM

Member

to MartinM
It's more about Jitter than ping. I have clients that connect from Charleston,SC. to San Jose (on a diff provider) with a consistent 115ms ping. ZERO problems. The provider is stable, in a high end DC. I support the Mega pop. I would rather have 5 great connections, than 10 'ok' ones, and not have to spend my time hunting down stable pops and configuring my DID's and end users to a new POP every couple of weeks when a flakey one decides to act up. Consolidate. (Auto failover would be nice, but if a mega pop alleviated lots of 'flakeyness', you shouldnt have to fail over as often..)
MartinM
VoIP.ms
Premium Member
join:2008-07-21

MartinM

Premium Member

We're not keeping more to have "ok ones".
taytong888
join:2005-06-20
Nepean, ON

taytong888 to MartinM

Member

to MartinM
Hello MartinM,

Vancouver is already served by Seattle. Why not have a robust POP in Calgary to serve the rest of British Columbia, as well as fast-growing Alberta, Saskatchewan and possible Manitoba?
Toollio
join:2003-11-17
Brazil/Cda

Toollio to chrisedwards

Member

to chrisedwards
said by chrisedwards:

This is still a fairly recent thread so I thought I'd add my thoughts:

1. I don't see a Europe POP mentioned. We have employees travel frequently to Europe and it's nice to have a European POP that they can reliably connect to while overseas. I haven't tried the New York POP though, so it's possible that in various locations in Europe the NY POP would have sufficient pings. In any case I'd love to see a European POP kept.

2. I like the New York POP selection. New York has good connectivity to South America. I get 100-120ms pings to Sao Paulo in Brazil. While higher than what'd you'd see domestically, it's workable.

3. I like Los Angeles POP which has good connectivity to Mexico (~60ms to Puerto Vallarta).

I agree that consideration should be given to customers outside North America. I'm in Salvador, Brazil, most of the time and I find that Montreal offers me one of the best ping times. In my case it beats New York and other U.S. servers. It might be prudent to do some testing to see whether the servers you intend to keep will satisfy customers in Europe, South America and elsewhere.

Latency tends to be overrated. I find up to 160ms perfectly acceptable, with no lag in conversation. Like the poster I've quoted, I am perfectly happy with 120, which is what I get from Salvador to Montreal. I've been using various VOIP services for years from Brazil and, of course, never get the kind of low latency numbers I get while in North America. But it's never been a problem. As somebody else points, out, jitter is another issue--bad jitter is a real problem.
MartinM
VoIP.ms
Premium Member
join:2008-07-21

MartinM to chrisedwards

Premium Member

to chrisedwards
said by chrisedwards:

This is still a fairly recent thread so I thought I'd add my thoughts:

1. I don't see a Europe POP mentioned. We have employees travel frequently to Europe and it's nice to have a European POP that they can reliably connect to while overseas. I haven't tried the New York POP though, so it's possible that in various locations in Europe the NY POP would have sufficient pings. In any case I'd love to see a European POP kept.

london.voip.ms is in Europe. The data center is close to London, and the bandwidth has direct peering to with London, England.

pende_tim
Premium Member
join:2004-01-04
Selbyville, DE

pende_tim to mozerd

Premium Member

to mozerd
+2 on auto-fail over
pende_tim

pende_tim to Blunderbuss

Premium Member

to Blunderbuss
The fallacy in the '100' and '200' year event comparison is that in NJ we have had 3 100 year events in the past 2 years!
drak0
join:2009-05-16

drak0 to MartinM

Member

to MartinM
I'm sad to hear that this plan has been put on the shelf. My wife uses voip.ms for her business line and Houston has for the most part been very good, but not having a graceful failover sucks... especially when I get a frantic call about the phones not working :/
JJ_GTA
Premium Member
join:2009-04-01
Ontario

JJ_GTA to MartinM

Premium Member

to MartinM
From a phone *user* point of view, I don't care what you do or how you do it. Just make the dial tone reliable. In the old day of POTS we didn't ask Bell how the infrastructure was built but we sure did depend on it. (Nobody knew when the phone was out of service unless we actually used it, compared to now, us knowing in an instant with alerts for failed registrations)

The challenge is how does my $5+ per month make a viable solution to design and build to make sure I can count on you for a quality call every time. I have to expect outages and failures, just not on a regular basis.
MartinM
VoIP.ms
Premium Member
join:2008-07-21

1 edit

MartinM to drak0

Premium Member

to drak0
said by drak0:

I'm sad to hear that this plan has been put on the shelf. My wife uses voip.ms for her business line and Houston has for the most part been very good, but not having a graceful failover sucks... especially when I get a frantic call about the phones not working :/

The study had nothing to do with the quality of service. We'll continue to push toward flawless QoS.

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

Davesnothere to SCADAGeo

Premium Member

to SCADAGeo
said by SCADAGeo:

I'll vote for any plan that increases reliability and yields the highest benefit-to-cost ratio.

 
+1

That about sums up MY feelings.
Davesnothere

Davesnothere to mozerd

Premium Member

to mozerd
said by mozerd:

+1

ESPECIALLY automatic fail-over between PoPs

 
And this too, of course.

Exact quantity of servers and their locations - YOU choose.

Jeffawa
@rogers.com

Jeffawa to MartinM

Anon

to MartinM
I would not mind a consolidation of servers. My concern is that I have tried both Montreal and Toronto servers and had problems. Since moving to Chicago I have had no problems. With consolidated servers my options for bad quality become limited.
w1ve
Premium Member
join:2007-12-28
Hancock, NH

w1ve to MartinM

Premium Member

to MartinM
I'm 100% in favor of it, and have been a customer for years.

I have a hosted PBX in NJ, ping time to NY 4mS. I used to host on thw west coast, and had ping times in the 2mS range. The POPs you are suggesting make perfect sense.