Atlanta, Florida, Virginia dedicated

[Todd Holley]

I am going to expand to one of those 3 locations.... I really would like to see your opinion as far as which one is best, and where I can get the machines at. I was pretty much thinking Virginia but Atlanta would also be great.

 

Thanks!

[Daersun]

As one of the most connected cities in the world, I like Atlanta, you can serve pretty much all of the SE from there.

[qwidjib0]

I would go with Atlanta next as well; and Jordan would be the expert on that.

 

Virginia is pretty close to New York. Florida does have a surprisingly big Internet scene, until you realize that it's really not intended for the United States, being separated all the way down there from the rest of the country. Most connectivity to South America leaves the Nap of the Americas in Miami via submarine fiber runs. South America's Internet itself is still very non-developed, so Florida is just where a healthy percentage of South American websites are hosted (all of those scandalous "off-shore hosted" websites excluded).

[jbiloh]

We offer Ashburn VA, but so long as you have a NYC presence don't bother with VA. For game servers, go with Seattle or San Jose before any othe locations you listed.

[DougK94]

Where are your targeted customers from? Since you have NYC already, most of the east coast is covered pretty well. Like Jon said, I would not waste time and $$ with VA. For customers in the SE, ATL will get lower pings than NYC, but is it enough to make that much difference?

 

With where you are now, you have east and central covered fairly well. I personally would go west before I went into ATL. Then after that, I might look at ATL again. Just my $ 0.02

[Todd Holley]

We only have a couple of western clients. Almost all of our clients are central / east. And a lot of them are in the square from dallas to atlanta to new york to chicago then back to dallas. But I may consider San Jose as we have gotten a few requests for it.

[Todd Holley]

We offer Ashburn VA, but so long as you have a NYC presence don't bother with VA. For game servers, go with Seattle or San Jose before any othe locations you listed.

 

Jon you don't offer Seattle do you?

[JasonF]

You can use Ubiquity server or Softlayer for Seattle.

[Todd Holley]

Softlayer? = Did NOT know they expanded to seattle. Very nice! Although I don't really like their services. I am pretty sure I am going with Jon (as always) in San Jose.

 

Thank you guys for your input. I appreciate it very much!

 

 

 

Edit: And apparently they are also expanding to New York or Chicago after a chat with one of the Sales guys.

[Daersun]

Id say stick with Jon in SJ - its a great location, beats the pants off SL

[Todd Holley]

Id say stick with Jon in SJ - its a great location, beats the pants off SL

 

Talking to Jon now.

 

If any one is looking for a great dedicated server host. Go CC, you will be happier than you have ever been. I would also like to thank this time to thank Jon for everything... continue the good work!

[qwidjib0]

http://www.softlayer.com/press_2007_12_17.html

 

SoftLayer customers will now be able to select the geographical location of their server deployment in real time, an industry first. “You can choose whether you server is installed in Dallas, Seattle, or first available, right as you’re ordering it. And the ability to choose your location is free of charge,” said Steven Canale, SoftLayer Vice President of Sales. “It’s part of our strategy to give customers more control.”

 

zomg! We had a good laugh about that one at our office when this came out. Being the first company to ever allow customers to choose their server location from a website is probably going to revolutionize the Internet.

 

Jon I didn't realize you were in San Jose and not Los Angeles, Our sales team can probably send you some nice leads.

[jbiloh]

http://www.softlayer.com/press_2007_12_17.html

 

 

 

zomg! We had a good laugh about that one at our office when this came out. Being the first company to ever allow customers to choose their server location from a website is probably going to revolutionize the Internet.

 

Jon I didn't realize you were in San Jose and not Los Angeles, Our sales team can probably send you some nice leads.

 

San Jose has already been real good to us, I can't complain. jbiloh@colocrossing.com is the email.

 

Perhaps some day we can show you the non-Mzima light in Chicago

[Defcon|Rich]

http://www.softlayer.com/press_2007_12_17.html

 

 

 

zomg! We had a good laugh about that one at our office when this came out. Being the first company to ever allow customers to choose their server location from a website is probably going to revolutionize the Internet.

 

I believe you misread what was being said. When you click submit on their order form your server begins the provisioning stage which is fully automated. The software does all the work and to my knowledge is not something anyone else has the capability to do thus far. We are talking about a multimillion dollar company here with resources me and you can only dream about..

 

Compare that to most of the other providers that offer 24+ hour setup times (on average) and rely on a $8/hr tech to manually load the OS and you will start seeing the benefit of automated provisioning..

[qwidjib0]

That may be, though that's not how they spelled it out above (and it wouldn't make any sense in line with "whichever is available the fastest" as an option). Yes Lance does have some big venture capitalists wrapped around his little finger; money is nice, though I can't imagine running a company in this industry that's mostly owned by people that are only expected to understand finance; most dedicated companies fall into that trap, and it hasn't worked out as well for all of them as it has so far for softlayer. Jordan didn't you and Ryan talk to him at hostingcon?

 

I don't know of any data center techs that get paid $8/hour (our data center techs sure get a good chunk of change more ), I don't believe that would even cover cost of living in most US cities. We've played with automated provisioning systems ourselves via PXE boot, it works pretty simply, though it isn't all that efficient unless you're stocking like them, or only offer a couple standard options with no customizations. If we have hardware on the data center floor servers are always loaded in under 24 hours though, the 24-72 hour waits are when we need to have something custom overnighted. Installing Linux takes less than 30 minutes and just a few simple steps; that's really not much to ask from a company with people on-site 24/7; even with fresh installs placed at the bottom of the order in priority.

 

ev1 did automation as well, as well as lots of companies that use NocMonkey and some other simple solutions. I heard stories of ev1's old refurbished hospital turned data center in Houston where they would pull the UPS trucks right down the aisles and slap down their white-box towers right there on breadracks off the back of the truck several times a day... certainly a different world.

[Defcon|Rich]

Cost of living has gone up that's for sure but I remember theplanet putting out job offers last year for entry level noc monkees starting at $8/hr. It would appear you get what you pay for though

 

That doesn't surprise me with EV1. They are numero uno in the walmart hosting market..

[qwidjib0]

Perhaps some day we can show you the non-Mzima light in Chicago

 

What for? Mzima has 400+ peers and a sFlow-powered intelligent routing, delivered from the only terabit routers in production (72.37.224.2 is Chicago). Then they take down the operating costs of this by having their own private fiber in the picture (in contrast to Internap). The only problem we've had is ancient folklore from when they first started saying it was only good in Los Angeles, when Los Angeles was their only POP. Now their office is in New York (not LA) and the map looks like this.

 

The only other network I've ever seen that's been consistently comparable has been Internap.. which we'll likely be adding as the secondary (and only viable secondary) provider in 2008, as it also uses intelligent routing via their proprietary FCP product and has strong peering, and wouldn't cheapen the network with a plain BGP counting hops.

[qwidjib0]

Cost of living has gone up that's for sure but I remember theplanet putting out job offers last year for entry level noc monkees starting at $8/hr. It would appear you get what you pay for though

 

Haha absolutely.

[jbiloh]

What for? Mzima has 400+ peers and a sFlow-powered intelligent routing, delivered from the only terabit routers in production (72.37.224.2 is Chicago). Then they take down the operating costs of this by having their own private fiber in the picture (in contrast to Internap). The only problem we've had is ancient folklore from when they first started saying it was only good in Los Angeles, when Los Angeles was their only POP. Now their office is in New York (not LA) and the map looks like this.

 

The only other network I've ever seen that's been consistently comparable has been Internap.. which we'll likely be adding as the secondary (and only viable secondary) provider in 2008, as it also uses intelligent routing via their proprietary FCP product and has strong peering, and wouldn't cheapen the network with a plain BGP counting hops.

 

Mzima has come a long way from the days of their scary-poor NYC network. If you're happy with your current setup, thats great. My offer was more sarcasm based than not - but hey, if you're interested, let me know!

[qwidjib0]

No worries.. I'm pretty sure we signed something that triggers an incredible hulk / lex luther reflex when they still get ragged on off the west coast (things posted on forums years and years ago can be tough to get out of people's minds). We do love Mzima in all of our locations; Chicago definitely no exception... it's actually our first/largest POP.

[Monk]

What for? Mzima has 400+ peers and a sFlow-powered intelligent routing, delivered from the only terabit routers in production (72.37.224.2 is Chicago). Then they take down the operating costs of this by having their own private fiber in the picture (in contrast to Internap). The only problem we've had is ancient folklore from when they first started saying it was only good in Los Angeles, when Los Angeles was their only POP. Now their office is in New York (not LA) and the map looks like this.

 

The only other network I've ever seen that's been consistently comparable has been Internap.. which we'll likely be adding as the secondary (and only viable secondary) provider in 2008, as it also uses intelligent routing via their proprietary FCP product and has strong peering, and wouldn't cheapen the network with a plain BGP counting hops.

 

Mzima doesn't have that many peers.. they rely mostly on tier1 transit, unlike other tier2s.. mzima leases their pipes from savvis/level3 if i remember correctly, or maybe it was someone else. regardless. sflow is neat, but it's foundry only, netflow is nice too, but it's not available on anything cost effective from cisco. Foundry has decent cost ratio per port, but their CLI used/is/was buggy. Cisco is the de-facto standard, but it can get expensive if you need routing gear to do full bgp + high PPS on a gig port.

 

It looks like mzima doesn't have AT&T anymore. Internap has alot of peers, considerbly more than mzima could ever have.

 

-M

[Daersun]

Jordan didn't you and Ryan talk to him at hostingcon?

 

Well between our two companies running around grabbing as many balls as we could, I did get a chance to talk to Lance - had some interesting things to say, but I disagree with alot of what he said.

[qwidjib0]

Mzima doesn't have that many peers.. they rely mostly on tier1 transit, unlike other tier2s.. mzima leases their pipes from savvis/level3 if i remember correctly, or maybe it was someone else. regardless. sflow is neat, but it's foundry only, netflow is nice too, but it's not available on anything cost effective from cisco. Foundry has decent cost ratio per port, but their CLI used/is/was buggy. Cisco is the de-facto standard, but it can get expensive if you need routing gear to do full bgp + high PPS on a gig port.

 

It looks like mzima doesn't have AT&T anymore. Internap has alot of peers, considerbly more than mzima could ever have.

 

-M

 

In all seriousness, I really am not allowed to disclose who Mzima's peers are, but I assure you that they absolutely have over 400 peering providers (which is extremely high, and includes most all high profile ISP's I could think of). You can check this number with them yourself if you care to do some real research, their phone number is +1 888.446.9462. I'd also love to see the documentation on Internap's peers that you must have, white papers would be swell.

 

Mzima's upstream is Savvis, Level3, SBC/AT&T, and their own fiber runs by the last documentation we were given, and performance at all 5 locations with Mzima has only improved, not diminished over the past 12 months. Whether they lease or own (or both) their fiber, I honestly don't know, but I don't know how it would be relevant, outside of a very, very far reach attempt at a smear by a colo cross employee.

 

Also, sFlow is proprietary to Force10, not Foundry. I'd take Force10 over Cisco any day, but those huge routers aren't cheap. The Force10 Terabit E-Series routers Mzima uses are the largest routers in production - much larger than anything Cisco builds; with massively more 10Gigabit uplink capacity, Mzima's the only carrier yet using those still pretty new beasts. sFlow isn't just a neat gimmick, it's a massively superior form of optimizing routes on a global scale when used the way that Mzima does. Traditional data center routing optimizes by counting hops to choose shortest routes from just one point; more than anything it's funny that such a plainly flawed system like that is still the norm.

[Monk]

I wasn't aware f10 had sflow on their linecards. I don't keep up with vendors we don't use (at all). I apologize.

 

Anyways.

 

route-server>sho ip bgp 72.37.224.2
BGP routing table entry for 72.37.224.0/21, version 521620
Paths: (18 available, best #13, table Default-IP-Routing-Table)
 Not advertised to any peer
 7018 3356 25973, (received & used)
   12.123.45.252 from 12.123.45.252 (12.123.45.252)
     Origin IGP, localpref 100, valid, external
     Community: 7018:5000 7018:31022
 7018 3356 25973, (received & used)
   12.123.13.241 from 12.123.13.241 (12.123.13.241)
     Origin IGP, localpref 100, valid, external
     Community: 7018:5000 7018:32104

 

I don't see direct peering between ATT and mzima, at least in ord.

 

Checking oregon-ix:

 

route-views.oregon-ix.net>sho ip bgp 72.37.224.2
BGP routing table entry for 72.37.224.0/21, version 7073008
Paths: (39 available, best #23, table Default-IP-Routing-Table)
 Not advertised to any peer
 2905 701 3356 25973
   196.7.106.245 from 196.7.106.245 (196.7.106.245)
     Origin IGP, metric 0, localpref 100, valid, external
 16150 25973
   217.75.96.60 from 217.75.96.60 (217.75.96.60)
     Origin IGP, metric 0, localpref 100, valid, external
     Community: 16150:63392 16150:65232 16150:65320
 6939 25973
   216.218.252.164 from 216.218.252.164 (216.218.252.164)
     Origin IGP, localpref 100, valid, external
 5459 25973
   195.66.232.239 from 195.66.232.239 (195.66.232.239)
     Origin IGP, localpref 100, valid, external
     Community: 5459:3 5459:60
 6461 25973
   64.125.0.137 from 64.125.0.137 (64.125.0.137)
     Origin IGP, metric 0, localpref 100, valid, external
     Community: 6461:5997
 6453 3356 25973
   207.45.223.244 from 207.45.223.244 (207.45.196.93)
     Origin IGP, localpref 100, valid, external
..

 

I only see 3356 (level3), and above (6461), I don't see savvis at all.

 

route-views.oregon-ix.net>sho ip bgp 72.37.224.2 | include 3561
 3561 25973
 1239 3561 25973

 

Ok, there's savvis. I think they might be prepending a couple of times to make 3356 more prefered.

 

Fixedorbit reports, initially, 76 peers total. I know that doesn't include private peering, because of proxy aggregation etc, but it's a far cry from their marketing 400 peers

 

http://fixedorbit.com/AS/25/AS25973.htm

 

CIDR Reports suggests a different view, less peers than fixed orbit, it would be more if they were using oregon-ix again for traffic views.

 

http://www.cidr-report.org/cgi-bin/as-report?as=AS25973&view=2.0

 

I don't see mzima on the equinix peering exchange as well. I also only see a couple of Tier1s, and I don't see any end user networks. End user networks are great to peer with, because you offset the cost of paying for transit costs. However, it's difficul to peer with anyone like that, unless you're on equinix peering exchange, or switch-and-datas peering fabric.

[Monk]

It should also be noted that I found this rather amusing:

 

http://www.cidr-report.org/cgi-bin/as-report?as=AS18508

 

They are a customer of mzima.