|
|
Built for Reliability
- Most of our equipment is co-located in a
telecommunications facility in San Jose, California. There
we enjoy the benefits of non-stop power, air conditioning, secure
access, remote hands and access to major fiber carriers.
We have four independent paths to the Internet.
- Our Network Operations Center (NOC) is in a second
location. We backup critical systems to systems at the NOC.
We also monitor and manage the network from the NOC.
- Circuits: T1 circuits from customer locations travel across the
bay area and enter our building on fiber. These circuits are
then delivered to us on channelized DS3, which then plug directly
into our router ports. We do not drop down to the copper pair
level, which greatly enhances reliability.
Also, we can provision multiple T1 on different DS3 hubs and different
routers.
- Routers: we run dual cisco7206vxr routers. We use HSRP to give
co-location customers that extra bit of redundancy and each
router has an independent path to the complete Internet.
- DNS: we run the industry standard
bind (named)
on
Solaris
and freebsd platforms.
We run different versions (when available), on different hardware and different OS to enhance
diversity and reduce the possibility that a single defect will affect
all nameservers. Our nameservers are in two datacenters.
One of the nameservers carries the "football"... but we won't say which one.
Besides, it changes frequently because you can never be too careful.
Besides our registered nameservers, we have other machines ready to
take over nameserver duties should the need arise.
Please note that we run two sets of independent machines for the two
independent parts of the Domain Name System: one set for our customer's
domains that we serve to the world, and the other set for our customer's
to use for general name queries (caching name servers).
- email: we run four mail servers that provide backup or primary
mail queue service. These machines provide anti-spam filtering.
- Accurate time: we run two network time protocol (ntp) servers.
- To increase redundancy, we use disk mirroring, raid, system
cloning, warm and cold spares, and extra socks. We have cold
spares off-site, should a catastrophe occur at the primary site.
All contents Copyright © 2007 The Internet MainStreet, Inc.
|
|
