| Posted: October 19, 2009, 6:01 pm - IP Logged | |
Please be aware that on Thursday, October 22, Lottery Post will be unavailable in the afternoon for a period of between one and two hours.
Unfortunately, I will not have the precise time when the site will become unavailable until right before it happens, so the best I can offer at this point is that it will be sometime after 1:00 p.m. Eastern Time.
Once the site is unreachable, it will be offline for between one and two hours. Naturally, I will do my best to finish up and get it back online as quickly as I can.
I'm mentioning this now (Monday) to provide as much advance notice as possible, so our members and visitors can plan accordingly. You should plan on positing predictions, blog posts, daily forum posts, etc., in the morning on Thursday, so that you are not interrupted on Thursday afternoon when the site becomes unavailable.
The maintenance is part of an upgrade in which I am moving the Lottery Post servers and other equipment to a brand new data center, complete will new power backup and redundancies that should prevent some of the nagging downtime issues we've faced a few times this year.
There are actually two independent power feeds to the building, with 2 utility transformers and an in-building switchgear room. The main feed surge supression is rated for "life services", similar to what a hospital would use. It is supposed to eliminate voltage irregularities and brown-outs. Automatic transfer to backup generator will continue feeding power in complete black-outs for 15 hours without refueling. The building has a 60 million volt lightning strike and grounding system.
Of course, all of the Lottery Post equipment will also be on its own independent UPS system, so even if a rare power spike were to happen, the battery and voltage regulation of the UPS would ensure it did not impact the servers.
Also, the network redundancies at the new facility are far superior to the old one (which were pretty good to start with). It houses a redundant routing infrastructure, redundant IPS, dual fiber entry points, a DWDM network that is connected diversely to 4+ IP carriers, a new DNS infrastructure, etc., etc.
Overall, reliability should be increased significantly, with fewer outages and a more reliable network.
Since I'm taking the time to work on the equipment, I'm also setting aside some time to work on the disaster recovery. In addition to the general backups that I have in place now, I plan on laying the groundwork for additional hot-swap backup servers to minimize the loss and amount of downtime in the event of complete system failure.
