Uptime

When "DOWN" does not mean "OFF": Uptime Explained
 

Since 2009 we have been working on various methods to mitigate downtime for our sites and services.  One of the techniques we have implemented most recently is the use of multiple "virtual" servers in various locations to ensure that one technological glitch does not bring down EVERYTHING associated with a given website.  For example, users are automatically balanced between at least two different web servers during peak periods, and email delivery issues do not impact our customers' ability to read existing mail and send out new messages.

We utilize a service called UPTIME ROBOT* to monitor various areas of our infrastructure, and these will be explained below.  Since service monitoring began in 2010, we have achieved over 98% uptime across all of our services.

Client websites

Each site is monitored in four locations, on at least two servers.  This provides fail-over and redundancy.  We also cache some of our sites' content through the CloudFlare CDN [content distribution network] and employ Secure Sockets Layer (SSL) as well as Round Robin DNS when possible, to ensure that momentary server issues do not affect access to websites.  Our client web servers also run MySQL database services; more on that below.

Domain Name System

Accessed and monitored from at least four geographical locations.  DNS is what translates web addresses, such as www.google.com into their Internet Protocol address (74.125.226.116 in the preceding example), which is what is actually used for server communications.  Our two "Cloudflare" servers are also monitored, and all DNS services fully support the IPv6 protocol, which is the next generation of Internet address space that was publicly rolled out in 2011. 

Mail Services

Three main services are monitored, IMAP and POP3 [used for receiving e-mail via a client like Outlook or our own webmail service] and SMTP [used for sending messages out to other users].  We also have a secondary POP3 "gateway" service which is used for importing popular "webmail" accounts into a mail client.  As of September, 2013, mailboxes and related administration is spread out over two servers, and the main web mail interface is on a third server.  All data is backed up daily, as with all of our other servers.  It should be noted that if the mail server is completely down, incoming mail is always queued for at least 7 days before a non-delivery report is sent back.

Proxy Services

A premium service we offer to our clients to bypass geographical restrictions their ISP has placed on them, or if our overseas clients need access into the United States.  It is important to note that this service is subject to prior approval, and does not fall under our regular downtime compensation policies.  This service is also subject to change or termination at any time, based on usage patterns and datacenter response.  Read more...

MySQL Databases

MySQL (Structured Query Language) databases are used when needed by our clients to access their websites.  Databases are used by programs like the WordPress blogging platform or other Content Management Systems [CMS].  We run local MySQL database servers on the same network as our clients' primary web servers, and maintain at least 2 remote/backup MySQL servers.

As of March 1, 2017, our home page has been updated with not only a running list of downtime and uptime reports, sorted by most recent date, but a link to an easy to read STATUS PAGE.    You may notice that even though a certain server or service shows that it is DOWN, your website or e-mail continues to operate normally.  That's the whole idea of this multi-faceted infrastructure; and while there may be false positives at times (monitoring is done every 5 minutes), changes to servers should be seamless and bugs should be unobtrusive.  Our Windows servers regularly go "down" for short periods in the early hours of the second Tuesday of the month for Microsoft updates, while our Linux servers check for software and kernel updates daily; this should be generally seamless as well.   Further, while the majority of the servers monitored have close to a 99.9% uptime, we only guarantee 25% uptime on each of our four (4) DNS servers, leading to over 100% uptime for the entire set.  Uptime on our main website, as well as specific client websites is also enhanced through the use of the two additional DNS servers on the Cloudflare CDN.  It is highly unlikely that every one of those six DNS servers would go down at the same time; as long as any two are up, all our hosted websites will likewise remain available.

When services cannot be automatically "failed over" to a working server, that 5 minute monitoring window allows us to react faster and perform a manual switch, or inform our customers with more accuracy as to the timeline of service restoration.  

* GIVING CREDIT: We would like to acknowledge the following individuals and companies which currently comprise our hosting infrastructure:

Alpha Geek, Coresite, Fibernetics, Global Telecom, Hurricane Electric, Incero, Joshua Anderson, MXGuardDog, OVH SAS, Quadranet, Sprintlink, Uptime Robot Bilisim A.S., Vimeo, Vultr, and Yomura Holdings.

We would also like to acknowledge Mr. Mark Brown of Triple 8 Networks in Los Angeles, and Mr. James Cornick of JCHost Networks in Kansas City for their assistance both past and present in virtual server deployment for our customers.