amuck-landowner

Search results

  1. D

    Colocrossing's Core "Router" is a Switch

    I lost all respect for cisco the day they threatened OpenBSD with a patent lawsuit over VRRP (aka CARP). Their products are overrated. 
  2. D

    Colocrossing's Core "Router" is a Switch

    Trust me, that is exactly what they were when the 720's first came out. We priced a 6506-E with 3bxl, 2 6724's and 2 copper gig blades, it was $39,000USD (I can probably find the email from my vendor).  We looked at the sup2's as well, due to their raw PPS (and competitive pricing), the sup32's...
  3. D

    Colocrossing's Core "Router" is a Switch

    As I was the principle person responsible for setting up the network during it's birth, I am going to go out on a limb and defend taking default routes. Routers/switches with large amounts of memory for routing are not cheap, years and years ago when the routing table was ~200k routes (with...
  4. D

    Need some advanced consulting?

    I'm looking for some temporary work, as the holiday season is up and coming. Do you  - Have questions OS design (Linux, FreeBSD, Solaris, NetBSD, OpenBSD, AIX, BSDi, et and Windows based systems) /debugging/troubleshooting/performance? - Need hardware advice?  - Have a webpage that is slow...
  5. D

    Web performance (E3 or E5)

    E5's have AES-NI. If you're doing lots of crypto (SSL), you can't go wrong.
  6. D

    SSL Speed Up?

    On SSL, Session init() is expensive due to socket calls, and entropy from SSL itself. Opening up /dev/urandom on any *NIX OS takes a very long time, due to stat64() call on /dev/random then an open(), then finally a read() Long ago apache had to disable KeepAlives/Pipelines on SSL connections...
  7. D

    Colocrossing's Core "Router" is a Switch

    We setup default routing long ago and did not take full tables since we were not really multihomed, years ago a sup720 costs an arm and a leg to do full BGP tables, so it was just more worthwhile to take defaults and use multihop with run-of-the-mill switching gear.  Giving users BGP isn't...
  8. D

    DNS Server Locations

    It's not bad, as a full lookup puts it's toll on the root servers. It's common because in RFC1034, section 4.3.4 states that a server /can/ do it, not that it must or must not.  Uncached: monk@deathstar:~$ time nslookup www.ais.net  Server: 127.0.1.1 Address: 127.0.1.1#53  ...
  9. D

    DNS Server Locations

    BIND with forwarders give you good resolution speed. Most networks cache DNS for about 20 minutes to an hour and let it expire. 
  10. D

    SSL Speed Up?

    What is 'enable hardware acceleration' ? How does one 'Cache SSL sessions' ? 
  11. D

    What kind of spec to handle 100k pageview/day

    I would avoid using APC because it does not page opcode to disk, so it stays in memory. Free memory on *NIX servers goes towards the filesystem buffer cache, generally. 
  12. D

    SSL Speed Up?

    You'll need a hardware crypto accelerator (PCi-E one) or a processor that supports AES-NI instructions to get better speed.  An earlier poster suggested these, I'll skip the protocols, but look at the cipher list:   ssl_ciphers...
  13. D

    What kind of spec to handle 100k pageview/day

    My thoughts: - There's a WP plugin to convert the dynamic pages into static HTML images, which reduces CPU load from the PHP processor. Then, you could easily configure apache/nginx/whatever to mmap the html files into memory to reduce CPU load (due to copy on write).. One downside to mmap is...
  14. D

    Apache 2.4 Mess

    Do you /really/ need Apache 2.4? Apache hasn't had the same Quality since the 1.3 days, it's been too politically influenced. Not only that, a good portion of their 'core team' is made up of people pushing their own political agenda. I found a race condition in Apache 2.0 some time ago relating...
  15. D

    The truth about ColoCrossing

    Yes, I did, for almost a decade, till May of last year.. 
  16. D

    The truth about ColoCrossing

    You really have no idea.....
  17. D

    ioping 0.7 does microseconds!?

    One of the 'cavets' to doing latency calculations on Linux with I/O is that, for example, ioping has the following section of code. (Linux has a shared ELF page at the top of the stack that caches lookups for time and a few other syscalls so you'll never be able to know the true amount>. Only...
  18. D

    The truth about ColoCrossing

    I've never spoken to that Chris person in my life, I have no idea who he is, what he does. I've heard of him, but never talked to him.  Summit was a company to take on HiDef and others; It was pretty good for the most part (this was so long ago, I was in my mid 20s when that company was...
  19. D

    The truth about ColoCrossing

    You have me at a disadvantage, what kind of interest has surrounded my name?  Gibson did not do anything in the game server field. He was an author of a VoIP program that was a total waste of money. No insult to him, he just wrote a program in a field he didn't understand. It had potential, but...
  20. D

    What are your RAM limits for different use cases?

    The problem with memory usage in VPS's is that the guest OS arch has a lot to do with memory usage. 64bit software uses way more memory than 32bit due to swollen pointers and alignment padding. When you use 64bit with say, 512MB of memory, a mmap() call will actually consume almost 3x the amount...
Top
amuck-landowner