Results 1 to 24 of 24
  1. #1

    Rackspace Network vs.....?

    I've posted a few posts in the forum so far and I'd like to thank everyone for their replies and advice.

    We are going to colocate our servers from out of rackspace to another location and I was wondering how the rackspace network compares to most other colo networks out there? Our main concerns are latency and uptime. We host a CRM application that is accessed by our clients 24/7 and they seem to notice even the "slighest" of slow downs.

    How does the network compare to the likes of Internap, Peer1, Level3/Abovenet, etc.

    Thanks again guys, you have all been supremely helpful.

  2. #2
    Join Date
    Feb 2004
    Location
    Atlanta, GA
    Posts
    5,662
    Well, you are comparing a provider's network to that of networks of individual carriers.

    Ultimately the most important thing first and foremost in your network is to have carrier redundancy. Under no circumstances would I ever personally place any equipment in a facility that only has connectivity from a single carrier.

    The next thing to look at is what redundancy is within the network. Is your potential provider competent in their network design? Have they designed a distributed network able to sustain failures? All too often you'll find smaller places (which it sounds like you aren't looking at) that will do something like take a pair of Cisco 6500's and connect their upstream connections as well as their downstream clients all in one box with no "layers" to the network, one fails, it all fails. That is a situation to be avoided.

    Rather than attempt to compare carriers (which rackspace buys from and has in their network mix) what are the providers (colo/network) you are evaluating? We would be able to give you much more valuable feedback if it was an apples to apples comparison

  3. #3
    Well, we are evaluating a few places actually.

    On our short list right now we have iland, peer1 and possibly phoenix nap. We'd like to stay in LA if possible, but its not a strict requirement.

  4. #4
    Join Date
    Oct 2003
    Location
    Hanoi
    Posts
    4,309
    If you don't want to stay in LA, I would highly suggest Netriver in Seattle. You will have fully redundant setup there. Contact Adam Vierra for more details.

  5. #5
    Join Date
    Aug 2003
    Location
    /dev/null
    Posts
    2,132
    1. Latency is solely relative to the place where you are located and the place where your customers are located.

    2. If it is really mission critical, and money is a secondary issue to you, I would try to avoid the facilities that sell to the end-user market (or the "average joe doe", like selling dedicated servers to gaming providers, shared web hosting companies, etc). Not because they are or are not good, but because the type of customer they have is more propense to have DDoS attacks and link saturation (as well as link overselling), port scanning, switch takedowns. If you are at rackspace, their overpricing sometimes is good because it protects their own network from cheap IRC kiddies and dumb shared hosting server administrators trying to mess things around and/or leaving their servers unsecured and open to attacks. Stick with Corporate-specialized colocation facilities.

    3. Avoid secondary facilities of a same company in a city. I.e., if the company you are seeking for has their main building at location X, and they have a nearby facility at location Y = X + 10 miles, avoid being in location Y. This because it most likely will be a single point of failure, as they would be probably backhauling all their carriers from facility X into facility Y by doing DWDM or similar transport technology, that may or may not be redundant and may or may not be oversold/overloaded (there are recent reports on the California area with some facilities, just dig up here at WHT Search tool).

    4. What is your total bandwidth requirement? This number would help if known, as we could recommend you going directly into two different Tier 1 carriers (especially selecting between those that don't share any kind of infrastructure in order not to have any single point of failure, i.e., Level 3 and Global Crossing share the ownership of some physical fiber lines, so having both and only both not always is the best deal), or could recommend you to go the Tier 2 way with good Tier 2 guys, or even building your own AS and doing your peering agreements on a public exchange.

    5. It would help also to know if your customers are all corporate and which network(s) they are in. I.e., if everybody is inside Cogent (as they are present in a bunch of corporate and office buildings) it would be logical to have a direct connection with Cogent, as well as Verizon Business, Sprint, etc. Maybe you could get this data from your customers (doing a poll with their tech areas) and see which carriers you should target more directly.

  6. #6
    Thanks for the response!

    Our bandwidth usage is actually very minimal as our CRM is DB driven and is all text and very few images. I believe right now we only use approx. 2-3mbs.

    Our clients are specialized in what they do, but they small businesses often on DSL/Cable connections. The top 3 cities they are located in are Los Angeles (mainly downtown), Dallas and New York. We do not have any overseas clients, and approximately 90% of them are in the US with the other 10% being in Canada.

    Edit: As for budget; we only need a half rack but would get a full rack if needed (or required). Our budget is around $1000-$1200 including bandwidth.
    Last edited by pacmantravis; 11-07-2009 at 10:02 PM.

  7. #7
    How does Colo4Dallas and their premium network rank with these providers?

  8. #8
    Join Date
    Aug 2003
    Location
    /dev/null
    Posts
    2,132
    If you are doing strictly colo (i.e., putting there your own equipment), then they are the best and most solid facility you could be in Dallas. AFAIK they don't sell dedicated servers (Colo4Paul can correct me on this).

  9. #9
    Yes, we're doing strictly colo.

    Depending on what sort of pricing we get from them, they definitely are on the list. I put in for a quote through the website and am waiting to hear back.

  10. #10
    Join Date
    Nov 2009
    Location
    Reston
    Posts
    305
    I would suggest doing an ASN search on fixedorbit dot com. This will allow you to view anyone's backbone and peers.

    Rackspace AS number is 27357 and colo4dallas is 30496 .... If you compare the two networks it would look like this:

    Colo4dallas peers and transit providers:


    Neighbors
    ASN Name
    3356 LEVEL3
    4323 TWTC
    12179 INTERNAP-2BLK
    20202 CRUCIAL
    23486 NETSPAN
    26774 TERRAPACKET
    32420 APTILO-US
    33210 1-800-HOSTING
    34146 ISP-SOLUTIONS-AS
    36352 AS-COLOCROSSING
    40610 DFW1
    47011 Unknown

    and rackspace peers and providers:
    ASN Name
    209 ASN-QWEST
    1299 TELIANET
    3356 LEVEL3
    6461 ABOVENET
    7018 ATT-INTERNET4
    10913 INTERNAP-BLK
    36248 RMH-14

    Which providers do you think are more valuable?

  11. #11
    Join Date
    Dec 2004
    Posts
    86
    The question here is really how many of these are transit providers and how many are not. On Colo4dallas it looks like 3 transits vs 6 transits for Rackspace.

    Quote Originally Posted by dbuyer0 View Post
    I would suggest doing an ASN search on fixedorbit dot com. This will allow you to view anyone's backbone and peers.

    Rackspace AS number is 27357 and colo4dallas is 30496 .... If you compare the two networks it would look like this:

    Colo4dallas peers and transit providers:


    Neighbors
    ASN Name
    3356 LEVEL3
    4323 TWTC
    12179 INTERNAP-2BLK
    20202 CRUCIAL
    23486 NETSPAN
    26774 TERRAPACKET
    32420 APTILO-US
    33210 1-800-HOSTING
    34146 ISP-SOLUTIONS-AS
    36352 AS-COLOCROSSING
    40610 DFW1
    47011 Unknown

    and rackspace peers and providers:
    ASN Name
    209 ASN-QWEST
    1299 TELIANET
    3356 LEVEL3
    6461 ABOVENET
    7018 ATT-INTERNET4
    10913 INTERNAP-BLK
    36248 RMH-14

    Which providers do you think are more valuable?
    ____________________________________________
    Brian W. Gemberling
    Pryme Technologies LLC
    http://www.pryme.net

  12. #12
    Join Date
    Mar 2007
    Posts
    402
    Quote Originally Posted by rackguru View Post
    The question here is really how many of these are transit providers and how many are not. On Colo4dallas it looks like 3 transits vs 6 transits for Rackspace.
    Also worth noting FixedOrbit is horribly out of date sometimes. Using Route Views or a similar server will give you much better BGP data.
    iCall Carrier Services - Carrier-grade VoIP services from a licensed CLEC - http://carriers.icall.com
    Domestic termination and origination, toll-free origination, A-Z International termination, dedicated servers, and colocation in our wholly-owned datacenter
    Real-time ordering via our control panel or XML-based API with over 20,000 numbers in stock

  13. #13
    Join Date
    Oct 2002
    Location
    Vancouver, B.C.
    Posts
    2,699
    Quote Originally Posted by iptelligent View Post
    3. Avoid secondary facilities of a same company in a city. I.e., if the company you are seeking for has their main building at location X, and they have a nearby facility at location Y = X + 10 miles, avoid being in location Y. This because it most likely will be a single point of failure, as they would be probably backhauling all their carriers from facility X into facility Y by doing DWDM or similar transport technology, that may or may not be redundant and may or may not be oversold/overloaded (there are recent reports on the California area with some facilities, just dig up here at WHT Search tool).
    Rather than just assuming there is no redundancy, this is something you should ask about. Although it's certainly true that some providers will simply have a single transport link with no redundancy, there are others who will have redundant links taking diverse paths and using different providers at their secondary location.

    Meanwhile, there are also providers out there who may have the connectivity of their primary facility depend on a single cross-connect. It's always better to ask rather than assume.


    Quote Originally Posted by pacmantravis View Post
    How does Colo4Dallas and their premium network rank with these providers?
    I'm not too familiar with iland or phoenix nap, but we colo with Peer 1 and I'd say they are different from most providers as they are also a carrier and own and operate their own network.

    My understanding of Colo4Dallas is that it is the typical data centre model, where they have various transits brought to a single location. Peer 1 has their own network with a full ring around North America and their own transatlantic links extending into Europe. They have tons of peers at the major peering exchanges in those regions, and have each of their upstreams available at most locations, resulting in much more flexibility for routing and optimization.
    ASTUTE INTERNET: Advanced, customized, and scalable solutions with AS54527 Premium Performance and Canadian Optimized Network (Level3, Shaw, CogecoPeer1, GTT/Tinet),
    AS63213 Cost Effective High Performance Network (Cogent, HE, GTT/Tinet)
    Dedicated Hosting, Colo, Bandwidth, and Fiber out of Vancouver, Seattle, LA, Toronto, NYC, and Miami

  14. #14

    I would recommend somebody in LA for this project

    Quote Originally Posted by pacmantravis View Post
    Thanks for the response!

    Our bandwidth usage is actually very minimal as our CRM is DB driven and is all text and very few images. I believe right now we only use approx. 2-3mbs.

    Our clients are specialized in what they do, but they small businesses often on DSL/Cable connections. The top 3 cities they are located in are Los Angeles (mainly downtown), Dallas and New York. We do not have any overseas clients, and approximately 90% of them are in the US with the other 10% being in Canada.

    Edit: As for budget; we only need a half rack but would get a full rack if needed (or required). Our budget is around $1000-$1200 including bandwidth.
    With this budget you'll be able easily to find a good colo provider (I know a lot of people are leaving Rackspace because of their pricing - something like $1000 for a manager server )

    Here, at webhostingtalk are a lot of budget companies with inexpensive network setup, Tier 2 providers (you saw that already in one of the posts in this thread), multilayer network gear starting with 65xx and going down to the cheapest one (vs having only high end equipment all the way through the network - you saw one post about or kind of about that too), who are the other customers is also important. Somebody may put you on 1 Gb/s network (I would recommend 10 Gb/s capable internal network) next to a potential "cheap" customer who can be under DDos attack - as a result the fragile network will go down and your servers along with that. A lot of details - do you want a one man company - CEO/Sales/CGO/Support guy, or you want a real company - you may not see all of that from the posts here, but I would recommend to consider that all and then plus some

  15. #15
    Join Date
    Jan 2003
    Location
    Chicago, IL
    Posts
    6,957
    Quote Originally Posted by WireSix View Post
    All too often you'll find smaller places (which it sounds like you aren't looking at) that will do something like take a pair of Cisco 6500's and connect their upstream connections as well as their downstream clients all in one box with no "layers" to the network, one fails, it all fails. That is a situation to be avoided.
    Overall, I agree with the intent of what you were saying, but the above part simply seems to be technically inaccurate. Putting everything on a pair of 6500's, upstream + customer VLANs, etc. can be done in a completely redundant manner. It is not a one fails, it all fails type configuration, at least not if it is properly configured. Saying that someone who has a pair of 6500 routers/switches doing all their core and distribution work is running a "one fails, it all fails" network configuration, is simply wrong.
    Karl Zimmerman - Founder & CEO of Steadfast
    VMware Virtual Data Center Platform

    karl @ steadfast.net - Sales/Support: 312-602-2689
    Cloud Hosting, Managed Dedicated Servers, Chicago Colocation, and New Jersey Colocation

  16. #16
    Quote Originally Posted by KarlZimmer View Post
    Overall, I agree with the intent of what you were saying, but the above part simply seems to be technically inaccurate. Putting everything on a pair of 6500's, upstream + customer VLANs, etc. can be done in a completely redundant manner. It is not a one fails, it all fails type configuration, at least not if it is properly configured. Saying that someone who has a pair of 6500 routers/switches doing all their core and distribution work is running a "one fails, it all fails" network configuration, is simply wrong.
    You are absolutely correct, not only the post was wrong, I don't think it was informative for the OP. But you know - posters are trying their "best" in posting their messages - sometimes it doesn't work out and then you are getting post like one above
    Professional Streaming services - http://www.tulix.com - info at tulix.com
    Double optimized - AS36820) network, best for live streaming/VoIP/gaming
    The best quality network - AS7219

  17. #17
    Join Date
    Feb 2004
    Location
    Atlanta, GA
    Posts
    5,662
    Quote Originally Posted by KarlZimmer View Post
    Overall, I agree with the intent of what you were saying, but the above part simply seems to be technically inaccurate. Putting everything on a pair of 6500's, upstream + customer VLANs, etc. can be done in a completely redundant manner. It is not a one fails, it all fails type configuration, at least not if it is properly configured. Saying that someone who has a pair of 6500 routers/switches doing all their core and distribution work is running a "one fails, it all fails" network configuration, is simply wrong.
    Karl,

    What I am referring to is a setup where it's that pair in the network and thats it. 100M access blades, upstream connections, everything in the pair. While it can be engineered for availability you run into issues with hsrp group # limitations, etc. With a network so concentrated at 2 points any individual failure has a greater potential for major impact. While with a higher device count in the network you may have a greater overall chance of a device failing the overall network impact is lessened.

  18. #18
    Quote Originally Posted by WireSix View Post
    Karl,

    What I am referring to is a setup where it's that pair in the network and thats it. 100M access blades, upstream connections, everything in the pair. While it can be engineered for availability you run into issues with hsrp group # limitations, etc. With a network so concentrated at 2 points any individual failure has a greater potential for major impact. While with a higher device count in the network you may have a greater overall chance of a device failing the overall network impact is lessened.
    Hey, Karl has more posts than you
    On a more serious note - 6509 to plug directly - 8x48 devices - 336 devices (assuming that you have redundant sup engines)
    35xx with 29xx hanging on them - approximately 576 devices (if switches are 24 ports).

    So one 35xx can take out more devices than one 6509 and plus in 6509 you have dual power supply and dual sup engines - I would go with the first config (based on 6509)!
    Professional Streaming services - http://www.tulix.com - info at tulix.com
    Double optimized - AS36820) network, best for live streaming/VoIP/gaming
    The best quality network - AS7219

  19. #19
    Quote Originally Posted by tulix View Post
    With this budget you'll be able easily to find a good colo provider (I know a lot of people are leaving Rackspace because of their pricing - something like $1000 for a manager server )
    Amen to that. Plus, since we want to virtualize...pricing on their "private managed clouds" makes it even higher. Nothing against then, they earn their money and are great at what they do. But we have a budget we need to stick to and there is no way we could do it with them while virtualizing our infrastructure.

  20. #20
    Join Date
    Nov 2009
    Location
    Reston
    Posts
    305
    Quote Originally Posted by KarlZimmer View Post
    Overall, I agree with the intent of what you were saying, but the above part simply seems to be technically inaccurate. Putting everything on a pair of 6500's, upstream + customer VLANs, etc. can be done in a completely redundant manner. It is not a one fails, it all fails type configuration, at least not if it is properly configured. Saying that someone who has a pair of 6500 routers/switches doing all their core and distribution work is running a "one fails, it all fails" network configuration, is simply wrong.
    Even by saying "a pair of 6500's" is implying redundancy! I agree with your comment. A pair of 6500's with VRRP and multiple providers sounds redundant to me!

  21. #21
    I think it is also important to understand how long it will take someone who is qualified to work on a router to be on-site. Some sites have spares of every card used and someone capable of swapping out any card should the need arise. Others may not be on-site but they have people contracted that are. I am not saying you can't do a good job without it, only that to really talk about redundancy you must also look at Mean Time to Repair if/when something fails. People are one of the most important factors in a real discussion of network planning and survival.

    As for peers vs transit you cannot look at a list of peers and say that one has higher performance than another. You dont know looking at a list whether a provider does basic BGP, route optimization or least cost routing. If more peers = higher quality then Internap would be one of the lowest quality provider in the last five years and Cogent would be one of the highest. By looking at a list of peers you do not know the type of connection (1Gbps, 10Gbps, multiple 10Gbps) or the percent of capacity they are.

    I see a lot of discussion here about redundant switches and really just shake my head. I am 100% in agreement that you need redundancy. We have multiple edge and core as many do. But it always frustrates me when we offer as a standard product to do HSRP to two different 6509s that very few people actually implement the dual link. Everyone talks redundancy for the carrier and how they wont go with someone that is not redundant, then they hook up one link instead of two.
    Looking for next opportunity

  22. #22
    We have servers colocated at Datapipe. We have been with them since 2000 i beleive. They have a rock solid network and we have only had one network outage in those 9 years.

  23. #23
    Join Date
    Jun 2001
    Location
    Denver, CO
    Posts
    3,302
    Quote Originally Posted by Colo4-Paul View Post
    I see a lot of discussion here about redundant switches and really just shake my head. I am 100% in agreement that you need redundancy. We have multiple edge and core as many do. But it always frustrates me when we offer as a standard product to do HSRP to two different 6509s that very few people actually implement the dual link. Everyone talks redundancy for the carrier and how they wont go with someone that is not redundant, then they hook up one link instead of two.
    And they typically also will not take A+B power, if and when it is available ...
    Jay Sudowski // Handy Networks LLC // Co-Founder & CTO
    AS30475 - Level(3), HE, Telia, XO and Cogent. Noction optimized network.
    Offering Dedicated Server and Colocation Hosting from our SSAE 16 SOC 2, Type 2 Certified Data Center.
    Current specials here. Check them out.

  24. #24
    Join Date
    Jan 2003
    Location
    Chicago, IL
    Posts
    6,957
    Quote Originally Posted by Colo4-Paul View Post
    I see a lot of discussion here about redundant switches and really just shake my head. I am 100% in agreement that you need redundancy. We have multiple edge and core as many do. But it always frustrates me when we offer as a standard product to do HSRP to two different 6509s that very few people actually implement the dual link. Everyone talks redundancy for the carrier and how they wont go with someone that is not redundant, then they hook up one link instead of two.
    Yep, we also offer HSRP/VRRP configs and have run into this many, many times... They complain that the HSRP/VRRP was supposed to prevent downtime, apparantly ignorant of the fact it was their single switch that failed, etc. They demand full redundancy, which we provide, yet don't spend the extra couple hundred dollars to make themselves fully redundant.

    As with A+B power, we have so few customers using our A+B feeds that at our new site we aren't even doing A+B feeds, though we'll still have some space left at 350 with A+B feeds if needed. I would think they'd at least want to dual cord core network gear, but many don't even do that.
    Last edited by KarlZimmer; 11-18-2009 at 11:48 AM.
    Karl Zimmerman - Founder & CEO of Steadfast
    VMware Virtual Data Center Platform

    karl @ steadfast.net - Sales/Support: 312-602-2689
    Cloud Hosting, Managed Dedicated Servers, Chicago Colocation, and New Jersey Colocation

Similar Threads

  1. RACKSPACE TEST: network unreachable!! ??
    By looking4dedicated in forum Dedicated Server
    Replies: 25
    Last Post: 04-30-2007, 05:59 AM
  2. Looking for Testers with our new Servers/ Control Panel. [RackSpace Network/ DC]
    By Ryan Smith in forum Other Web Hosting Related Offers
    Replies: 3
    Last Post: 05-10-2006, 06:50 AM
  3. Colo - Quality Network 10Mbps and up $60/Mbps $15 per U rackspace
    By RSanders in forum Dedicated Hosting Offers
    Replies: 4
    Last Post: 06-22-2004, 04:14 PM
  4. VDI CPanel3 and RackSpace Network. The best of both worlds
    By Ironlung in forum Shared Hosting Offers
    Replies: 0
    Last Post: 12-08-2000, 03:45 AM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •