Design

Design

Sorted By Creation Time

Web site availability

Contact:http://www.packetnexus.com

4 January 2001 Best Practices for Web Site Availability
A Gartner analyst offers advice on avoiding web site down time. Continuous
availability requires a "multi-pronged strategy"
incorporating application and infrastructure design, testing, and proactive
management. Two Internet company executives also offer their
list of best practices for availability.
Back to the Index

Web site availability

Contact:http://www.packetnexus.com

4 January 2001 Best Practices for Web Site Availability
A Gartner analyst offers advice on avoiding web site down time. Continuous
availability requires a "multi-pronged strategy"
incorporating application and infrastructure design, testing, and proactive
management. Two Internet company executives also offer their list of best
practices for availability.

[Editor's (Murray) Note: Not very rigorous. I am glad that they promote
business and people issues, limit single points of failure, and promote
up-front design but is that really their "best?" Where does it say it say
fix responsibility and accountability, limit complexity and
function, manage change, monitor, measure, report, etc. I admit that this is
a difficult area to write in but I would expect better.]

Back to the Index

Web site availability

Contact:http://www.packetnexus.com

4 January 2001 Best Practices for Web Site Availability
A Gartner analyst offers advice on avoiding web site down time. Continuous
availability requires a "multi-pronged strategy"
incorporating application and infrastructure design, testing, and proactive
management. Two Internet company executives also offer their list of best
practices for availability.

[Editor's (Murray) Note: Not very rigorous. I am glad that they promote
business and people issues, limit single points of failure, and promote
up-front design but is that really their "best?" Where does it say it say
fix responsibility and accountability, limit complexity and
function, manage change, monitor, measure, report, etc. I admit that this is
a difficult area to write in but I would expect better.]

link


Back to the Index

Server load balancing solution

Contact:http://www.packetnexus.com

         Either should be okay. A few things to think about

         NAT or not to NAT. Both will do it. Alteon was a bit easier to do
and seemed to work better, but I don't think I have all the kinks out of
the config on the Arrow yet.

         Arrowpoint supports ssh, older Alteon's don't. not sure about the
new stuff. I like the config interface on the Arrow better. It's more Cisco
like. I like the label system better on Arrow. You can create your own
names as opposed to having to remember which number you set www8 as.

         Number of machines. On the Alteon you only get 256... I think.
This may have changed depending on the code. 5.2.x is evil. 6.0.x is nice.
8.0.x has had problems according to the old FreeI admins who tested it for
their mail system. They were running 6.0.x and NATing on an Ace 3 and
thought it was fairly stable.

         . The Arrowpoint has a few more features and was a bit more
flexible for port redirections. Better healthchecks unless Alteon has
changed recently. I haven't kept up with the Alteon since we've been
migrating to the Arrows.

         We're currently running two sets of the css 11801's for http type
stuff and an Alteon Ace 3 for mail. The Ace 3 has been stable for us doing
mail once we moved to 6.0.x code. I forget the number of transactions per
second, but it's a solid 20Mb/s of mail during the day.

Ramin K
Network Engineer
Netzero Inc.

At 12:37 AM 12/20/00 +0100, Rod Oliver wrote:
>Hi Guys,
>
>We are in the process of evaluating server load balancing solutions for a
>high density mail system. We are looking into Alteon, Cat6000 and Cat4000
>series and the old Arrowpoint range. I realise that this is a kind of
>'horses for courses' question, but I would very much appreciate comments,
>anecdotal and technical that people might consider to be appropriate. The
>options that I have listed are mainly so because we are up to now a mostly
>Cisco shop.
>
>Rod Oliver


Back to the Index

Security and switches

Contact:http://www.packetnexus.com

I brought this up when we were talking about re-designing the CoLo network.
Here is someone else saying the same thing.

In a nutshell....Using VLAN's on a single switch for the entire network is
not a good idea and susceptible to attack and single point of failure.

We didn't buy that 6509 already, did we?

http://www.sans.org/infosecFAQ/switch_security.htm

jas


Back to the Index

Scalable Web Mail

Contact:http://www.packetnexus.com

 http://www.horde.org/papers/Scalable_webmail_HOWTO.php


Back to the Index

DNS

Contact:http://www.packetnexus.com

http://www.stamey.nu/DNS/Default.asp


Back to the Index

comp.protocols.tcp-ip.domains Frequently Asked Questions

Contact:http://www.packetnexus.com

http://www.intac.com/~cdp/cptd-faq/


Back to the Index