On Sun, 2009-03-15 at 11:23 -0500, Brad Knowles wrote:
> on 3/15/09 6:09 AM, Robert Hajime Lanning said:
> 
> > Can't read it. Not a member. Sorry.  I'll see if I can get to it in
> > October, when it is available.
> 
> I'll see if I can find the PDF version I have, and send that to you.

That would be cool.  Thanks.

> 
> > Anycast is not a widely used load-balancing technique.  As it requires
> > multiple geographically dispersed datacenters.  It is not used within
> > server farms, which most people are familiar with.
> 
> Actually, in some places it is used within server farms.  The ISC 
> approach (described at 
> <http://ftp.isc.org/isc/pubs/tn/isc-tn-2003-1.html>) is not the only one.
> 
> In practice, anycast has been successfully applied to DNS, as 
> f.root-servers.net has over 50 public installations around the world, 
> and probably at least a hundred more private ones.  Most anyone can get 
> their own private copy of f.root-servers.net, if they ask.  All of those 
> machines answer for the same IP addresses.  In theory, this should cause 
> problems with TCP-based connections, when routes change, etc....  In 
> practice that has not shown to be an issue.

So?  And how many people work in that environment?  Yes, it is there,
and a lot of people use the system.  But, it is invisible to the users.
Look at how many people manage load-balanced server farms.  And how many
people in that group, use anycast?  You are going to find the percentage
quite small.

Usually when you do use anycast, it is on-top of a standard
NAT/next-hop/proxy based cluster.  (And to be clear, I said "Usually"
not "Every".)

> 
> However, even though both protocols use UDP, NTP != DNS.  With NTP, 
> there's a lot more state stored by the client with regards to their 
> servers, and it's stored for long periods of time -- days, weeks, 
> months, years, or more.  If you can give me a 100% guarantee that there 
> will never, ever be any route flaps over that period of time, then maybe 
> this isn't an issue for you.  But that doesn't mean that this is a best 
> practice that others should follow, and it doesn't mean that you should 
> be encouraging others to do the same.

No, but if someone wants to do it, I would say "go ahead, but here are
the caveats..."  It might work enough for there needs.

One of the things I had learned at Seagate, was that the business
requirements need things to be "done."  My head kept having the
requirement of "done right."  I had to meet in the middle.

> 
> > And how many 1000 node LDAP server clusters have you seen?  (ie. the
> > original question was putting a few LDAP servers behind a load-balancer
> > for HA)
> 
> I run the LDAP servers for the uTexas Enterprise Directory system (see 
> <http://www.utexas.edu/its/ted/>), and we have three production LDAP 
> servers behind a pair of Netscaler fail-over load-balancing switches. 
> We have had issues with some of the servers getting out of sync.
> 
> > Seagate runs a suite of 5 read-only LDAP servers for their hole
> > corporation behind an F5 Big-IP load-balancer. (With a mirror on the
> > other side of the globe.)  There is a suite of 3 writable LDAP server
> > behind the same Big-IP, though different pool.  This runs their identity
> > management for the company.
> 
> We're working on re-architecting the system to try to reduce or 
> eliminate the synchronization problems, but just because your systems 
> have always worked perfectly over their entire history doesn't mean that 
> others haven't had significant issues.
> 
> > It is not hard.
> 
> Depends on how you try to change the one engine for the plane while it's 
> in the air.

Sounds like you are attempting realtime replication.  That is very hard.
Mainly because it is impossible to get 0 latency.

> 
> > Right, and who said I wanted to tell Howard Chu or Quanah Parker
> > anything?  I did not say it was wrong to point to the list.  I just did
> > not like the blanket statement that THE list was THE ONLY place.
> > 
> > It really comes across like you are saying that the people on the LOPSA
> > list don't know anything.
> 
> No, I wouldn't say they don't know anything.
> 
> But although I am subscribed to this list, and I've been supporting the 
> NTP Public Services Project for over five years and I have some 
> experience I can share with people, this still is not the appropriate 
> place to discuss NTP.  However, I can tell you where the appropriate 
> place is, over on the mailing lists I help run at lists.ntp.org.
> 
> Although I have been supporting the Mailman project, a member of the 
> Mailman steering committee, and the primary person running the mail 
> systems for python.org for over five years, does not mean that this is 
> the appropriate place to discuss Mailman.  Again, I can tell you where 
> the appropriate mailing lists are (for which I am the primary active 
> listowner), over on mail.python.org.
> 
> I see no difference for LDAP.
> 

Ok, so what is the purpose of the LOPSA "Tech" mailing list?  Since it
seems that all Tech topics have their own lists.  <sarcasm>I guess we
should disband this list.  And, everyone should go a subscribe to the
150 lists pertaining to all the topics of interest.</sarcasm>

These topics are not outside the list's stated topic.  Just because
there is a "better" place to discuss a specific topic, does not make
this an inappropriate place to also discuss it.  It is inappropriate
when it becomes off-topic for the list.

Now, I am going to stop here, as this has gotten off-topic for this
list.  (Also, I am starting a new job tomorrow, so I won't have time
to continue down the philosophical rabbit hole.  I was just bored this
weekend.)

-- 
END OF LINE
      --MCP

_______________________________________________
Tech mailing list
[email protected]
http://lopsa.org/cgi-bin/mailman/listinfo/tech
This list provided by the League of Professional System Administrators
 http://lopsa.org/

Reply via email to