Page 1 of 2 12 LastLast
Results 1 to 40 of 54
  1. #1

    Peer1, Peering or Reselling? vs Internap

    I've been told that Peer1 is just as reliable as Internap, but cheaper.

    My understanding of Internap from their site is that they purchase transit from many of the top carriers and resell a single connection to these many carriers using a proprietary BGP method that "finds the fastest route".

    My understanding of Peer1 is that they do the same using a less complex BGP setup.

    Am I understanding this correctly or is one of them actually a carrier that merely peers from other carriers and thus if a problem exists in their network you're gonna have some happy downtime?


    I'm looking for 100-200mbps unmetered bandwidth colo'ed in Seattle. Uptime is a must unfortunately and I don't want SLA uptime promises, I actually want uptime! It seems like with Peer1 I might be able to get into the reasonable price range of 30-40 per mbps (through a reseller most likely) but that with Internap it wouldn't even matter what the commit level was, I'd be paying out the bum. I am in no rush so I'm making sure I do my homework before making any purchases. Thank you in advance to any response and especially actual price discussion/comparison.

  2. #2
    Join Date
    Nov 2002
    Posts
    2,780

  3. #3
    Join Date
    Apr 2001
    Location
    St. Louis, MO
    Posts
    2,508
    If uptime is very important to you, Internap is going to be a good performer.
    Mike @ Xiolink.com
    http://www.xiolink.com 1-877-4-XIOLINK
    Advanced Managed Microsoft Hosting
    "Your data... always within reach"

  4. #4

    Re: Peer1, Peering or Reselling? vs Internap

    i realize you were probably looking for an answer from someone besides myself, but given that im the chief peer1 evangelist around these parts, here goes.

    Originally posted by HighCommit

    My understanding of Peer1 is that they do the same using a less complex BGP setup.
    this is correct. whereas internap relies on automatic performance measurement and dynamic route selection, peer1 does it the old way - by hand. internap has slightly better performance on some long haul routes as a result of that. due to peer1's strongly developed peering relationships, peer1 comes out slightly on top in certain geographic areas.


    Am I understanding this correctly or is one of them actually a carrier that merely peers from other carriers and thus if a problem exists in their network you're gonna have some happy downtime?
    the two are not opposites. you are understanding the nature of the offerings correctly. however, there seems to be a misconception wrt peering versus transit.

    both peering and transit are connections to a different network, differentiated only by the settlement arrangements (transit is paid, while peering is usually settlement free) and the range of routes presented in the view (you pay for connectivity to the whole internet with transit, while peering gives you access to the network in question).

    regardless of whether the connection is peering or transit, if the network you are connecting to has issues, those will be passed on to you. the whole purpose of having several transit providers (as both internap and peer1 do) is to have the ability to reroute traffic along a different vector, if you will, should issues crop up. problems with your transit provider are arguably more damaging than problems with your peer, since you typically connect to a much larger chunk of the 'net through your transit provider.

    peering relationships are Good (tm). why would you want your packets to get to broadband provider X through Y's network when you can send them to broadband provider X directly, shortening the route, which typically increases performance? you would not. peering improves performance and decreases cost. peer1, as its name implies, places a lot of emphasis on bilateral peering agreements. we have seen excellent performance which is directly related to their presence in nyiix and ash. since you are interested in seattle, peer1 participates at seattleix (six) as well.


    I'm looking for 100-200mbps unmetered bandwidth colo'ed in Seattle. Uptime is a must unfortunately and I don't want SLA uptime promises, I actually want uptime! It seems like with Peer1 I might be able to get into the reasonable price range of 30-40 per mbps (through a reseller most likely) but that with Internap it wouldn't even matter what the commit level was, I'd be paying out the bum. I am in no rush so I'm making sure I do my homework before making any purchases. Thank you in advance to any response and especially actual price discussion/comparison.
    internap is like every other carrier - if you get to the right person at the right time with the right amount of cash in your hand, you are going to get a good deal (where good deal is a good discount vis a vis list pricing). the trouble is that internap's pricing is quite high (compared to other carriers, disregarding performance) even if you get a great deal on pricing. whether the rather large difference in price is justifiable given the advantages (a few ms here and there on long haul routes) really depends on your application.

    i am naturally biased since we do a lot of business with peer1, but for what it is worth, i consider internap to be a good choice for applications where the profit margin is not significantly dependent on bandwidth costs. ie, if you are a large bank pushing a few hundred megs, you could care less what your bandwidth costs you (it is likely to be a small percentage of your it infrastructure maintenance costs). if you run a very successful ecommerce site, you are in pretty much the same boat. however, if you run a streaming media company or a game server provider, your bandwidth utilization is closely tied to your revenues (through the number of users) and bandwidth pricing is likely to have a significant impact on your bottom line. it is those applications that demand reliability and good performance, yet require good pricing as well, that i find peer1 to be well-suited for.

    paul
    * Rusko Enterprises LLC - Upgrade to 100% uptime today!
    * Premium NYC collocation and custom dedicated servers
    call 1-877-MY-RUSKO or paul [at] rusko.us

    dedicated servers, collocation, load balanced and high availability clusters

  5. #5
    Originally posted by RackMy.com
    If uptime is very important to you, Internap is going to be a good performer.
    cant argue with that, at least if you go by past performance. however, i am a bit concerned with their new AS-per-city topology. i recall an insightful comment on that topic posted in one of the xSP fora i frequent (could have been wht, nanog-l or isp-*), explaining the topology change in terms of decreased resilience to traffic volume based ddos attacks. i am not aware of an event substantiating (or refuting) that assessment, but it did make sense from an ip engineering point of view.

    paul
    * Rusko Enterprises LLC - Upgrade to 100% uptime today!
    * Premium NYC collocation and custom dedicated servers
    call 1-877-MY-RUSKO or paul [at] rusko.us

    dedicated servers, collocation, load balanced and high availability clusters

  6. #6
    Join Date
    Nov 2002
    Posts
    2,780
    You may also considered commiting to 2 Carrier with 100Mbps each and then run BGP on your own. If u're the kind of people that thinks you can hold fate in your hands, then you could control the UPtime by doing so.

  7. #7
    Originally posted by Mfjp
    You may also considered commiting to 2 Carrier with 100Mbps each and then run BGP on your own. If u're the kind of people that thinks you can hold fate in your hands, then you could control the UPtime by doing so.
    my guess is that he is (wisely so) trying to avoid the capex required to run bgp on a hardware-redundant setup. i wouldnt dream of running bgp on underpowered or non-redundant hardware in an 'uptime counts a lot' situation, and buying/operating two M5s is probably not the best thing one can do to stay within a tight budget.

    paul
    * Rusko Enterprises LLC - Upgrade to 100% uptime today!
    * Premium NYC collocation and custom dedicated servers
    call 1-877-MY-RUSKO or paul [at] rusko.us

    dedicated servers, collocation, load balanced and high availability clusters

  8. #8
    It doesn't seem like there's much to argue. You're turning me into a peer1 fanboy, and there's nobody here willing to bash them...

    not even a single person out there who had a bad experience with a peer1 phone tech1 causing you to go into a whirlwind of "peer1 sucks" posts?

  9. #9
    Join Date
    Oct 2003
    Location
    New Bedford, MA
    Posts
    79
    It will be hard to find someone who does not like Peer1.... That is not to say that there isn't someone out there

    They are a rock solid network. They are always a PLEASURE to work with both on the phone and via E-Mail.

    I would give them two thumbs up!

  10. #10
    Originally posted by HighCommit
    It doesn't seem like there's much to argue. You're turning me into a peer1 fanboy, and there's nobody here willing to bash them...

    not even a single person out there who had a bad experience with a peer1 phone tech1 causing you to go into a whirlwind of "peer1 sucks" posts?
    their noc technicians are very polite, so no. remember that they are a carrier, so you can not expect their remote hands folks to be able to restore a broken software raid setup by editing config files without proper supervision. with that said, we use our own techs, so this is a non-issue.

    paul
    * Rusko Enterprises LLC - Upgrade to 100% uptime today!
    * Premium NYC collocation and custom dedicated servers
    call 1-877-MY-RUSKO or paul [at] rusko.us

    dedicated servers, collocation, load balanced and high availability clusters

  11. #11
    Join Date
    Mar 2004
    Posts
    54
    Well, their San Jose site is kinda iffy. It's all peering and transit through only MCI/Worldcom. I can kinda understand that SJC is a new site for them and all, but hopefully they'll add some more transit soon.

  12. #12
    Originally posted by unavaca
    Well, their San Jose site is kinda iffy. It's all peering and transit through only MCI/Worldcom. I can kinda understand that SJC is a new site for them and all, but hopefully they'll add some more transit soon.
    dont forget that they can backhaul to any of their other pops if they need to route around a uunet problem. i also want to clarify that this is specific to san jose, which is their newest location.

    paul
    * Rusko Enterprises LLC - Upgrade to 100% uptime today!
    * Premium NYC collocation and custom dedicated servers
    call 1-877-MY-RUSKO or paul [at] rusko.us

    dedicated servers, collocation, load balanced and high availability clusters

  13. #13
    From what i understand (although Paul probably said it better than i could)- if you want top performance and don't care about the price, internap is a good option. However if you are going for value Peer1 is a LOT more affordable and the difference in performance between the two is not very much. This comes from some research i did comparing the two among others a little while back. Internap seems to only really make a latency difference on long routes such as going all the way across the country.

  14. #14
    Join Date
    Feb 2003
    Location
    Bay Area, CA
    Posts
    215
    If you want the best, go with InterNAP. If you are willing to settle for less than go another route.

  15. #15
    Originally posted by Huminie
    If you want the best, go with InterNAP. If you are willing to settle for less than go another route.
    thank you for your factual, insightful comment. you are an asset to this community.

    paul
    * Rusko Enterprises LLC - Upgrade to 100% uptime today!
    * Premium NYC collocation and custom dedicated servers
    call 1-877-MY-RUSKO or paul [at] rusko.us

    dedicated servers, collocation, load balanced and high availability clusters

  16. #16
    Join Date
    Feb 2003
    Location
    Bay Area, CA
    Posts
    215
    You are welcome to my opinion any time I choose to provide it to you.

  17. #17
    I have a few machines in Peer1 New York and find few faults with their network or servcie people. I have even had rare problems with the sales people, and if you don't mind me saying so, sales people for backbones are usually annoying sewer rats.

    Louis J Landerman
    USGameServers.com

  18. #18
    Join Date
    Mar 2003
    Location
    New Jersey
    Posts
    1,277
    just to clarify, peer1 doesn't really do anything different from any other providers (NAC, nLayer) like this. The only difference is the 100% SLA. I'm not saying they arent good, but they arent really as special as everyone sounds, just a large tier2 like the rest of 'em.

    Most big players actually optimize routing to specific endpoint ASNs that are more latent/lossy than the automagically selected route.

  19. #19
    Join Date
    Dec 2003
    Location
    Andromeda, DQ3XR13
    Posts
    159
    Originally posted by rusko
    my guess is that he is (wisely so) trying to avoid the capex required to run bgp on a hardware-redundant setup. i wouldnt dream of running bgp on underpowered or non-redundant hardware in an 'uptime counts a lot' situation, and buying/operating two M5s is probably not the best thing one can do to stay within a tight budget.

    paul

    Paul, how 'bout Zebra? Three *nix boxes running Zebra with one as a backup could provide pretty great redundancy couldn't it? That'd be a lot cheaper than M5's, and would offer similar performance?

    Also - Globix seems to have more peering relationship than Peer1, both in New York metro and in the world overall. Why doesn't anyone around here seem to recommend them? I'm guessing they have oversold their network or are otherwise inferior? From memory, they were the "original" Peer1 in NYC as far as putting a lot of energy into garnering hundreds of peering relationships instead of buying a lot of transit.

    Note that I've found in my own little tests over the years that Globix and NAC, despite having tons of peering, aren't quite as low-latency as Peer1 is. To me, Peer1>Globix>NAC... but all three seem fine providers 99% of the time. Would you tend to agree?

  20. #20
    Join Date
    Dec 2003
    Location
    Andromeda, DQ3XR13
    Posts
    159
    Originally posted by VAFS.COM
    To me, Peer1>Globix>NAC...
    Also, I'm aware that the WHT consensus seems to be Peer1>NAC>Globix, since NAC is everywhere here and Globix is never deemed worthy of mention. Is it that Globix is truly a worse network than NAC, or do they not offer competitive prices anymore? (They were pretty competitive a few years back in the days of Exodus owning the market.)

  21. #21
    Join Date
    Feb 2004
    Location
    Louisville, Kentucky
    Posts
    1,083
    Originally posted by VAFS.COM
    Paul, how 'bout Zebra? Three *nix boxes running Zebra with one as a backup could provide pretty great redundancy couldn't it? That'd be a lot cheaper than M5's, and would offer similar performance?
    My guess is that you don't have any experience deploying Juniper boxes, or an understanding of their product architecture, and that's why you believe PCs running Zebra would perform similarly to an M5. This is absolutely not the case.
    Jeff at Innovative Network Concepts / 212-981-0607 x8579 / AIM: jeffsw6
    Expert IP network consultation and operation at affordable rates
    95th Percentile Explained Rate-Limiting on Cisco IOS switches

  22. #22
    Originally posted by crnc
    just to clarify, peer1 doesn't really do anything different from any other providers (NAC, nLayer) like this. The only difference is the 100% SLA. I'm not saying they arent good, but they arent really as special as everyone sounds, just a large tier2 like the rest of 'em.
    mostly agreed. not all, even large, tier 2s have the appropriate levels of network redundancy at all of their pops. it isn't just about the SLA - the infrastructure to live up to it has to be there. as far as being special, i've argued against *any* provider being 'special' for quite a while now. people should evaluate carriers based on sane metrics. i like peer1 because it delivers connectivity me and my clients love. i'm sure richard and his clients can say the same.


    Most big players actually optimize routing to specific endpoint ASNs that are more latent/lossy than the automagically selected route.
    you meant 'less', i hope? <g>

    p
    * Rusko Enterprises LLC - Upgrade to 100% uptime today!
    * Premium NYC collocation and custom dedicated servers
    call 1-877-MY-RUSKO or paul [at] rusko.us

    dedicated servers, collocation, load balanced and high availability clusters

  23. #23
    Originally posted by VAFS.COM
    Paul, how 'bout Zebra? Three *nix boxes running Zebra with one as a backup could provide pretty great redundancy couldn't it? That'd be a lot cheaper than M5's, and would offer similar performance?
    you have got to be kidding me. if asics could be replaced with a few thousand lines of c code, don't you think things would be looking *very* different in the marketplace?


    Also - Globix seems to have more peering relationship than Peer1, both in New York metro and in the world overall.
    says who? i remember from some other thread that you were not familiar with the difference between bgp queries and looking at fixedorbit - therein lies your problem, i would think. besides, more is not always better - i care about peering to networks i have traffic going to in significant amounts.


    Why doesn't anyone around here seem to recommend them? I'm guessing they have oversold their network or are otherwise inferior? From memory, they were the "original" Peer1 in NYC as far as putting a lot of energy into garnering hundreds of peering relationships instead of buying a lot of transit.
    i am not too familiar with them, so can't comment directly. we currently host a couple folks that moved to us from globix and i've heard about a few nasty administrative issues.


    Note that I've found in my own little tests over the years that Globix and NAC, despite having tons of peering, aren't quite as low-latency as Peer1 is. To me, Peer1>Globix>NAC... but all three seem fine providers 99% of the time. Would you tend to agree?
    i would be interested in finding out what kind of testing you are doing, pm is fine. although i do agree, both peer1 and nac are fine providers. no comment on globix.

    paul
    * Rusko Enterprises LLC - Upgrade to 100% uptime today!
    * Premium NYC collocation and custom dedicated servers
    call 1-877-MY-RUSKO or paul [at] rusko.us

    dedicated servers, collocation, load balanced and high availability clusters

  24. #24
    Originally posted by VAFS.COM
    Also, I'm aware that the WHT consensus seems to be Peer1>NAC>Globix, since NAC is everywhere here and Globix is never deemed worthy of mention. Is it that Globix is truly a worse network than NAC, or do they not offer competitive prices anymore? (They were pretty competitive a few years back in the days of Exodus owning the market.)
    i'm not sure where you got your 'consensus' from. not sure about globix's pricing, but my bet would be on not competitive. the reason certain networks are mentioned often here is that well-repuated providers use them.

    paul
    * Rusko Enterprises LLC - Upgrade to 100% uptime today!
    * Premium NYC collocation and custom dedicated servers
    call 1-877-MY-RUSKO or paul [at] rusko.us

    dedicated servers, collocation, load balanced and high availability clusters

  25. #25
    Join Date
    May 2003
    Location
    Kirkland, WA
    Posts
    4,448

  26. #26
    Join Date
    Dec 2003
    Location
    Andromeda, DQ3XR13
    Posts
    159
    Originally posted by jsw6
    My guess is that you don't have any experience deploying Juniper boxes, or an understanding of their product architecture, and that's why you believe PCs running Zebra would perform similarly to an M5. This is absolutely not the case.
    You've guessed well, oh wise one. I'm actually looking for some specifics here though, not fanboy stuff. One Zebra box costing three figures and running FreeBSD can push through a full Gig-E with 0% CPU usage. What would the M5 do better?

    I'm actually trying to learn here, not flame... so please don't condescend. In the past, I've simply bought Cisco crap and let someone else manage it. In the future, it's very likely that I will buy an M5 (because I admit I've seen everywhere that a M5 is much better than the other commercial products on the market in that range) and let the telco manage my BGP. BUT, perhaps only because they won't manage Zebra boxes... (and also, M5's are not really that expensive anymore thank God).

    Anyways, thanks for any response.

  27. #27
    Join Date
    Dec 2003
    Location
    Andromeda, DQ3XR13
    Posts
    159
    Originally posted by rusko
    i'm not sure where you got your 'consensus' from. not sure about globix's pricing, but my bet would be on not competitive. the reason certain networks are mentioned often here is that well-repuated providers use them.

    paul
    Well, I'm not sure why it isn't obvious to you how I see the consensus. How many people have you seen on WHT who talk up Globix all day? None. How many people have you seen claiming that NAC is better than Peer1? None. (Some would say "as good as" though.) It doesn't take much observation of these threads to find a "consensus". Maybe if you'd like to dissect it semantically, I should say "vast majority opinion".

  28. #28
    Join Date
    Dec 2003
    Location
    Andromeda, DQ3XR13
    Posts
    159
    Originally posted by rusko
    you have got to be kidding me. if asics could be replaced with a few thousand lines of c code, don't you think things would be looking *very* different in the marketplace?
    But, I do see a rapidly changing marketplace. Zebra is finding its way into the game with *quite a few* hosting companies, and I can't imagine that the growth will slow down anytime soon. It's a good time to short Cisco (thanks to both Zebra's future and Juniper's present).

    Originally posted by rusko
    says who? i remember from some other thread that you were not familiar with the difference between bgp queries and looking at fixedorbit - therein lies your problem, i would think. besides, more is not always better - i care about peering to networks i have traffic going to in significant amounts.
    Well, can you tell me different? Peer1 doesn't even peer with UUNET or Level3 in NY (it seems to go through Savvis), and Fixed Orbit does list Globix with many more peering relationships. Is Fixed Orbit lying, or is Peer1 not announcing hundreds of peering agreements? Also, the link you provided didn't work yesterday or I would have checked it out.


    Originally posted by rusko
    i am not too familiar with them, so can't comment directly. we currently host a couple folks that moved to us from globix and i've heard about a few nasty administrative issues.

    i would be interested in finding out what kind of testing you are doing, pm is fine. although i do agree, both peer1 and nac are fine providers. no comment on globix.

    paul
    The little testing I do consists of running traceroutes and pings from various locations for random 24 hour periods and then seeing how much average latency and particularly *variance* I can pick up in the ping times. I find (and you may disagree?) that there is little more important in hosting than 1-uptime and 2-ping times (and IMO, variance).

  29. #29
    Join Date
    Dec 2003
    Location
    Andromeda, DQ3XR13
    Posts
    159
    Ok, I guess technically Globix doesn't fit into any consensus. Maybe that's along the lines of what you were trying to say... if no one knows about them, there's little opinion or consensus about them. Makes sense to me.


  30. #30
    Originally posted by VAFS.COM
    You've guessed well, oh wise one. I'm actually looking for some specifics here though, not fanboy stuff. One Zebra box costing three figures and running FreeBSD can push through a full Gig-E with 0% CPU usage. What would the M5 do better?
    the question is, gig-e of what? routers care about pps, not (just) throughput.

    paul
    * Rusko Enterprises LLC - Upgrade to 100% uptime today!
    * Premium NYC collocation and custom dedicated servers
    call 1-877-MY-RUSKO or paul [at] rusko.us

    dedicated servers, collocation, load balanced and high availability clusters

  31. #31
    Originally posted by VAFS.COM
    Well, I'm not sure why it isn't obvious to you how I see the consensus. How many people have you seen on WHT who talk up Globix all day? None. How many people have you seen claiming that NAC is better than Peer1? None. (Some would say "as good as" though.) It doesn't take much observation of these threads to find a "consensus". Maybe if you'd like to dissect it semantically, I should say "vast majority opinion".
    providers are tools/means to an end; as such, 'better' is a misnomer. for different people and/or applications, better may be as fast as possible, as reliable as possible, as cheap as possible or any combination thereof. as such, i try very hard not to get into comparison wars. it is impossible to keep those civil, so let's not start =]

    paul
    * Rusko Enterprises LLC - Upgrade to 100% uptime today!
    * Premium NYC collocation and custom dedicated servers
    call 1-877-MY-RUSKO or paul [at] rusko.us

    dedicated servers, collocation, load balanced and high availability clusters

  32. #32
    Join Date
    Dec 2003
    Location
    Andromeda, DQ3XR13
    Posts
    159
    Internet traffic... same packets as web hosting companies generally face. Mind you that I have not set this up myself, but have seen here and especially in Zebra mailing lists that this is true. I've not set one up myself to test because I don't even know my way around Zebra very well yet... but I'd like to learn it if it would be useful to me. (I suppose it would be at least useful to "learn" generalities about routing and BGP management that would be applicable to Juniper/Cisco/Zebra and then one day... ONE DAY.... maybe... MAYBE... I could save some money by doing it myself. But that's doubtful.)

  33. #33
    Join Date
    Dec 2003
    Location
    Andromeda, DQ3XR13
    Posts
    159
    Originally posted by rusko
    providers are tools/means to an end; as such, 'better' is a misnomer. for different people and/or applications, better may be as fast as possible, as reliable as possible, as cheap as possible or any combination thereof. as such, i try very hard not to get into comparison wars. it is impossible to keep those civil, so let's not start =]

    paul

    Ahhh... okay. Very good point. I'll steer clear of that then.

  34. #34
    Join Date
    Feb 2004
    Location
    Louisville, Kentucky
    Posts
    1,083
    Originally posted by VAFS.COM
    You've guessed well, oh wise one. I'm actually looking for some specifics here though, not fanboy stuff. One Zebra box costing three figures and running FreeBSD can push through a full Gig-E with 0% CPU usage. What would the M5 do better?
    CPU utilization is a non-issue on Juniper boxes. The M-series boxes do 2.5Gb/s bi-directional forwarding on each FPC, with line-rate firewall rule evaluation. It's route policy evaluation is also more advanced than what is offered by Zebra or IOS.
    Jeff at Innovative Network Concepts / 212-981-0607 x8579 / AIM: jeffsw6
    Expert IP network consultation and operation at affordable rates
    95th Percentile Explained Rate-Limiting on Cisco IOS switches

  35. #35
    Join Date
    Dec 2003
    Location
    Andromeda, DQ3XR13
    Posts
    159
    Originally posted by jsw6
    CPU utilization is a non-issue on Juniper boxes. The M-series boxes do 2.5Gb/s bi-directional forwarding on each FPC, with line-rate firewall rule evaluation. It's route policy evaluation is also more advanced than what is offered by Zebra or IOS.
    Thanks Jeff! Tell me more about the route policy evaluation... does this give it ...

    A) Faster routes
    or
    B) More reliability

    Also about the firewall rule evaluation, are you saying that this is a big factor if you are using a ...

    A) hardware firewall behind your router
    B) software firewall behind your router
    or
    C) the M5 as a firewall itself

    Thanks again, this is a big help.

  36. #36
    Originally posted by VAFS.COM
    But, I do see a rapidly changing marketplace. Zebra is finding its way into the game with *quite a few* hosting companies, and I can't imagine that the growth will slow down anytime soon. It's a good time to short Cisco (thanks to both Zebra's future and Juniper's present).
    the keyword being 'hosting companies'; ie not carriers and not providers. softrouters have their uses, but not at the core/edge with high pps and five nines requirements (imo).


    Well, can you tell me different? Peer1 doesn't even peer with UUNET or Level3 in NY (it seems to go through Savvis)
    the peering relationships you should be looking at are with eyeballs-heavy carriers. besides, the performance of savvis' transit is nothing short of stellar. as for why that peering is not in place, peer1 is uunet's customer (elsewhere) and the big tier 1s will in general not peer with tier 2s under these conditions.


    , and Fixed Orbit does list Globix with many more peering relationships.
    once again; more is not better. fixedorbit does *not* use bgp data, it uses the irr data. use netlantis (when it is up again) or show ip bgp.


    Is Fixed Orbit lying, or is Peer1 not announcing hundreds of peering agreements?
    fiuxedorbit doesn't claim to provide complete routing data, thus can not be lying. if peer1 were not announcing their routes, no traffic would be going over them. if the question is whether peer1 does not list all of their routes to peers in the irr, the answer is yes. registering route objects in the irr is voluntary and is only needed for things like prefix filter list generation.


    Also, the link you provided didn't work yesterday or I would have checked it out.
    what link?


    The little testing I do consists of running traceroutes and pings from various locations for random 24 hour periods and then seeing how much average latency and particularly *variance* I can pick up in the ping times. I find (and you may disagree?) that there is little more important in hosting than 1-uptime and 2-ping times (and IMO, variance).
    jitter (you call it variance) is only important for certain applications, such as streaming and gaming, where the data is both realtime and sequence sensitive. as far as pinging, i hope you weren't pinging routers as cef would throw you off just like that =]

    p
    * Rusko Enterprises LLC - Upgrade to 100% uptime today!
    * Premium NYC collocation and custom dedicated servers
    call 1-877-MY-RUSKO or paul [at] rusko.us

    dedicated servers, collocation, load balanced and high availability clusters

  37. #37
    Originally posted by VAFS.COM
    Internet traffic... same packets as web hosting companies generally face. Mind you that I have not set this up myself, but have seen here and especially in Zebra mailing lists that this is true. I've not set one up myself to test because I don't even know my way around Zebra very well yet... but I'd like to learn it if it would be useful to me. (I suppose it would be at least useful to "learn" generalities about routing and BGP management that would be applicable to Juniper/Cisco/Zebra and then one day... ONE DAY.... maybe... MAYBE... I could save some money by doing it myself. But that's doubtful.)
    for that matter, you could have an extreme blackdiamond do your bgp. of course, faced with high ppf it will spiral to a spectacular death, killing your uptime. you said uptime was important - ability to handle abnormal conditions, which are not that infrequent by the way, is the cardinal requirement for reliability.

    paul
    * Rusko Enterprises LLC - Upgrade to 100% uptime today!
    * Premium NYC collocation and custom dedicated servers
    call 1-877-MY-RUSKO or paul [at] rusko.us

    dedicated servers, collocation, load balanced and high availability clusters

  38. #38
    Join Date
    May 2003
    Location
    Kirkland, WA
    Posts
    4,448
    Hey now..lay off the Extreme hate.

  39. #39
    gg nick. i don't hate extremes, they are excellent for l2. flow-based routing just doesn't cut it in the environments i care about though.

    paul
    * Rusko Enterprises LLC - Upgrade to 100% uptime today!
    * Premium NYC collocation and custom dedicated servers
    call 1-877-MY-RUSKO or paul [at] rusko.us

    dedicated servers, collocation, load balanced and high availability clusters

  40. #40
    Join Date
    Dec 2003
    Location
    Andromeda, DQ3XR13
    Posts
    159
    Paul or other, are there any routers made by anyone not named Juniper that are similar to the M5 in price and function (and have the same "wirespeed" characteristics that Juniper is famous for)?

    [Someone stop me if this is getting too far off-topic.]

Page 1 of 2 12 LastLast

Posting Permissions

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