Client Area Login
6 SW 13th St, Ste 2
College Place, WA 99324
Obsessive network engineering at your
Visitors to your web site will simply choose a
different web site to view rather than wait for your web site to function. Your web site
must always be accessible and at a level of performance such that no visitor spends time
waiting on your web hosting company's network to deliver the data.
To create and maintain a network that delivers
100% of the time at levels of performance that exceed all expectations requires a special
kind of obsession. Let us handle the "obsessing over the network" part.
It's what we do and we love it.
Here is our special obsession:
Dual redundant Gigabit Ethernet links over a city-wide redundant fiber optic ring to our primary backbone provider. A secondary backbone provider uses a third Gigabit Ethernet link that enters the building at a different point of entry.
The upstream primary backbone provider's fiber optic network is fully redundant with transparent fail-over in the event of a fiber cut. The secondary backbone provider utilizes the local telephone company fiber optic network, which has very little down time.
- The upstream primary and secondary backbone providers' IP networks are fully BGP multi-homed with multiple high bandwidth links to multiple carriers. They also directly peer with many regional and national networks at Seattle area network exchanges, Spokane area network exchanges, and Portland (Oregon) area network exchanges.
- Enough fiber optic capacity for nearly limitless bandwidth. With present technology, there is nearly 8000Gbits/sec of bandwidth available on the fiber optics at our facility or,
in other words, enough capacity to transmit 232 full length (2 hours) high definition (1080p) movies per second. Of course, this immense capacity is only turned on incrementally as required.
- Cisco 7206VXR with NPE-G2, 2048MB of RAM, and dual redundant
load sharing power supplies. It has three gigabit Ethernet ports (1000Mbit/sec each) and numerous other lower bandwidth ports. If needed, additional ports can be added. This
high-end router was originally placed into service in 2001 when it was brand new. The
processing core was upgraded on January 9, 2008 with brand new components that give it
approximately 10 times the processing power it had when it was originally placed into
service in 2001. This router has been meticulously maintained over the years and we
anticipate it will have many more years of valuable service. We have several spare chassis, power supplies, processor cards, and port adapters on hand to enable quick repairs, should they ever be needed. We closely monitor CPU utilization, link error rates, and link utilization and will upgrade as needed.
- Cisco 7513 with dual redundant auto-failover route switch
processor cards and dual redundant auto-failover power
supplies. This router contains many high bandwidth ports. There is a tremendous amount of available expansion left in this router
and it will easily scale to multi-gigabit speeds if required. This router is currently on cold-standby and can be easily deployed as needs dictate. It will be retired in the near future and replaced by a much more power efficient unit with significantly higher network capacity.
The routers are connected to our LAN through
an auto-failover redundant fast Ethernet mesh comprised of multiple switches and multiple
100Mbit/sec switched links to protect in the event of any switch or port failure. The mesh
is configured in such a way that it can handle up to two failed switches without losing
connectivity between the routers and the LAN. This redundant fast Ethernet mesh is in the
process of being upgraded to gigabit Ethernet with a 48 gigabit redundant ring topology.
Transparent firewalling and bandwidth management system. This keeps unwanted traffic off the network and blocks known hacker
networks. Dynamic attack detection and firewalling automatically blocks the bad guys. The firewalling and bandwidth management system can handle up to 8 gigabits per second of aggregate traffic and can scale to 80 gigabits per second with upgrades.
Managed gigabit Ethernet switches
with a 48 gigabit redundant ring interconnect. Each server gets a dedicated 1000Mbit or
100Mbit port to our network, depending on the device. Over 600 gigabits of non-blocking
capacity and ultra low latency. For special needs we can provide switched 10 gigabit
Ethernet ports or bonded ports.
Data backup system
Backup and restore staging area on hard
disks at an off-site location. This ensures that file searches, backups, and restores all happen very quickly. The very
fast seek times of hard drives also allow us to do one full backup each month with daily
incremental backups without incurring unreasonably long restore times. It also enables a
smaller storage requirement for backups than if we did full backups every day. This system
is completely extensible and will be expanded and upgraded as required.
For systems that use our data backup system,
backups occur daily between 4PM and 6AM Pacific time. This is the period in which backups
are least likely to impact performance for most customers. Dedicated server and colocation clients can work us to specify a schedule to meet their needs if they have purchased the daily off-site backups service.
- APC Symmetra LX fully fault tolerant site-wide battery backup
system. This battery backup system is fully scalable and all critical components are n+1
redundant. It self-tests all components every two weeks and sends us an email that goes to
two cell phones if any problems are found during the self-test or during
normal operation. It also sends an email to two cell phones if any parameter
such as temperature or voltage is abnormal. Just about every possible aspect of the system
is monitored continuously and logged every 5 minutes. In the event of a power outage, we
have over 30 minutes of run-time on battery. This gives us plenty of time to prepare our
generators for power outages over 30 minutes.
We have two computer-grade gasoline generators
continuously on-site, which is enough to power our entire network and all servers with
power to spare. We keep enough fuel on hand for about 10 hours of continuous running. That
gives us plenty of time to procure more fuel should there ever be a power outage longer
than 10 hours. The longest power outage we've seen in our area between 1995 and now was
about 5 hours.
A live full power backup system test, including battery backup, generators, cabling, and power switching equipment, is done monthly to verify readiness. In the rare event an issue is identified, it is fixed as soon as possible and the entire system is retested.
Due to the geography of our area (in a valley
surrounded by mountains on three sides), any power disturbances are localized and not over
widespread areas like you'd find after a major hurricane or other widespread disaster.
This helps keep the mean time to repair relatively low for electrical outages in our area.
Furthermore, we are just a few blocks from the local power company's repair truck dispatch
facility and in the high priority repair airport area.
Temperature and humidity in our datacenter is
continuously monitored for abnormal fluctuations. If any problem is detected, our
environmental monitoring system pages two cell phones so that
the problem can be investigated and corrected.
Datacenter is cleaned weekly to remove any
dust that may have accumulated with normal use.
Security system is monitored 24/7.
Datacenter access is only by escort from one
of our staff.
Due to our proximity to the regional airport,
our area is constantly patrolled by airport police. If our security system detects an
intrusion, they are typically on-site within one to two minutes of receiving the call.
Situated at the foot of the Blue mountains in
the southeast corner of Washington state in Walla Walla, the geography of our location
naturally has a lower risk for natural disasters such as floods, tornadoes, earthquakes,
tsunamis, mud slides, hurricanes, and volcanoes versus the typical datacenter on one of
the coasts or in southern states. The area is in a valley surrounded on three sides by
Our location within Walla Walla is at the
regional airport, which was formerly a military airbase. It sits at a higher elevation
than the rest of the valley and thus avoids any flood risk from the nearby rivers and
creeks. When selecting this location for an airbase, the military naturally took into
account those risks.
The northwestern quadrant of the United States
has several active and dormant volcanoes. Though our location is not directly at risk from
any of those volcanoes, ash fall from any of those volcanoes could be a concern. To deal
with this risk, we have taken steps to prevent fine volcanic ash from reaching and
damaging our equipment, including electrical generators and datacenter air conditioning.
We also have an ash mitigation plan that allows us to continue uninterrupted operations in
the days and weeks of clean-up that could follow an ash fall event.
There are no known risks of floods, tornadoes,
earthquakes, tsunamis, mud slides, or hurricanes at our location.
In the event of a fire or other event that
destroys servers, data backups are done over the network to an off-site location.