The Magic of Hosting

Clarke certainly got a lot of it right in focusing on advanced technology. But that's not the entire equation when it comes to providing world-class web hosting. After all, even though high-end servers make for a great hosting experience, those servers are only as good as the team that supports them. Empowered hosting comes about only when sufficiently advanced technology is combined with sufficiently impressive support staff, administrative staff, and management. Of course, all of that has to be combined in the most optimal way possible. And, it has to work efficiently too — particularly given the time critical nature of ensuring reliable performance and high server uptimes. Growth plays a role too — when server farms grow from tens to hundreds to thousands of servers, whole new complexities get added. If you guessed that we sometimes wish everything was, really truly "magic", you're absolutely right! Fortunately, though, sufficiently advanced technology behind the scenes — along with a healthy dose of planning, common sense, hard work, dedication, and some simpler technologies, too — can more than compensate for a lack of magic. In case you're still not entirely convinced, we decided to break it down, and give everyone a bit of an insider's view of Net Logistics.

One of the cornerstones of providing effective customer service is to make oneself available to their customers regardless of the time of day, all the while considering the customer's preferred way of communicating. Although many people like e-mail, it's undeniable that sometimes talking to someone, person-to-person, is just as good of a medium (and perhaps even better!) for communicating concerns, seeking information, or buying a product or service. This is why we offer 24-hour telephone support.

There's not really a whole lot of "magic" to making such a system work well — it's mainly common sense and, yes, advanced technology. Standard VoIP technologies provide the backbone of our telephone support system. A very talented Australian voice artist provides the friendly voice behind our telephone menu (and rumour has it she's also the voice behind a rhinoceros on a popular BBC children's television show). Features within our internal support ticket system also ensure that company actions are quickly generated, scheduled and assigned with appropriate urgency, based on incoming calls.

Common sense and people power, though, form the true cornerstone of telephone support. This takes the form of a combination of rotating shifts, training of all staff in providing effective, friendly telephone support, and the expectation that everyone — even management — will make "front line" customer service their top priority. In other words, we make possible what would otherwise be impossible with advanced technology alone.

Of course, many customers prefer the convenience of e-mail based customer support, rather than (or in addition to) phone support. Shocking as it may be, the insanely fast response times within this system are not the product of time machines, time distortion, or any other form of magic — they're simply a result of an efficient, total system working as it's supposed to.

In terms of advanced technology, that is an important component of our e-mail customer service as well — we use a customised version of a popular support ticket system, called Kayako. Integrated into this are components of our proprietary Hawkeye Server Analysis, Metrics and Monitoring system. It is the latter system — Hawkeye — that proves critical in providing our support teams with the ability to engage in proactive support as much as possible, versus reactive support. In other words, it enables us to catch problems before they become major issues, and in the overwhelming majority of incidents, before they have an opportunity to adversely affect our clients' uptime/service availability. The ticket system also enables us to very effectively maintain various priority queues for support action requests, ensuring that issues are quickly escalated based on administrator requirements and issue urgency.

Ensuring questions, concerns and issues are effectively handled can't be entirely left up to technology, however. This is why our staff is also intensely trained in providing e-mail based support, just as they are with telephone customer service.

Naturally, a hosting environment just doesn't make a lot of sense without servers, underlying network infrastructure, and a reliable physical hosting location. Just as with impressive support times, impressive server performance isn't anything magical — hardware-wise, our servers are entirely "stock" — but of the very high-end, Dell® variety. It's worth mentioning here that some have asked why we choose to utilise Dell servers, rather than other brands. The simple answer is that although there are, in fact, many firms that offer good servers, Dell has the best combination of industry-acceptance, reliability, hardware support, and underlying cost. In other words, they truly offer the best "bang-for-our-buck," and thus enable us to effectively achieve our service quality objectives as well as passing along cost savings to our customers. A related question is why we don't use "beige-box" servers — i.e. standard desktop computer systems. The reasons for this are primarily based on reliability; rackmount servers are specially designed to deal with intense use and the issues that arise from that, such as component heat/required cooling. Other benefits include the ability to use standard server racks (rather than specialised equipment), which in turn enables more efficient use of datacenter resources.

Not everything involves sticking to standards, though, particularly in the smaller details. For example, all of our server wiring is, literally, engineered — that is to say, all network and power cabling systems are custom-designed by a professional engineer. As simple as it sounds, it is very dramatic in terms of making datacenter operations far more efficient; not only does it look tidy, it makes it very easy to do tasks such as moving servers between racks and installing cabling for new servers, and ensures one doesn't have to second guess what cable goes where! Then there's the matter of actually keeping track of equipment — this actually requires a fair bit of resources and of itself, and a mix of custom and standard software — considering that we have servers, routers, and other hardware spread amongst a variety of facilities both in Australia and abroad.

Speaking of facilities — that is another very critical component of maintaining excellent service levels. Although we discuss our network and datacenter infrastructure elsewhere, it is worth addressing a few of their similarities here. When we have made choices to expand our service offerings, we have been extremely selective in choosing datacenters. There are a range of criteria that rank as extremely important; although it's difficult to put them in an order of priority (since they are all truly critical to maintaining high quality of service), among them are highly redundant power systems, intense physical security, advanced fire suppression systems, adequate cooling, and well-engineered facility design. The reason for this is almost every one of those criteria simultaneously lowers risks of data loss and/or downtime, which in turns allows us to ensure higher levels of service.

Network infrastructure is very important as well. Although uplink quality is very important, one area that many people often overlook is internal network quality. In other words, one can have the very best and fastest uplinks on earth, but if the underlying routing and switching technologies that connect the web servers to this uplink are in any way inferior, overall hosting performance will be degraded. This is why we go out of our way to ensure that even our fail-safes have fail-safes; we only utilise enterprise-grade Cisco routers and switches, ensure those are configured with automatic fail-over characteristics for the rare cases in which those may fail. Additional redundant routers are also kept at hand, to ensure that we can quickly bring our network back online even in the case of a catastrophic failure of multiple routing components. Although it may seem odd, contingency plans exist even beyond that point.

Most of the above discussion has involved hardware — but there's a bit more to the apparent "magic." If you guessed that it's software, well, that's absolutely right! Over and above the "standard" software we use for providing product support (e.g. Kayako for the ticket system, Asterisk for our VoIP platform, etc.), there is a variety of additional, specialised software that helps make Net Logistics operate effectively and efficiently. One of the most important components in our operations is virtualisation software; namely, Parallels' Virtuozzo. Not only does Virtuozzo make our VPS plans possible, it also provides the basis for DDS — dynamic dedicated server — technology. To make a long story short, we essentially deploy dedicated servers on our network as DDS servers — VPSes that happen to run one VPS instance on each dedicated server (or "node"). Once that is accomplished, a whole new range of previously unavailable features comes to the forefront. In addition to the server management capabilities of Virtuozzo, DDS adds the ability for our dedicated servers to be extremely rapidly "moved" to a totally different physical server. In other words, by nature of virtualising our dedicated servers, we are able to make them almost completely independent of their physical hardware — a major benefit for purposes of backups, server upgrades, and server failure-risk mitigation. Although powerful commercial software provides some excellent technologies (that may even seem magical!), a lot of our operations also depends upon software we have developed in-house. This includes our custom-designed corporate intranet, our hardware inventory tracking system, and perhaps most notable of all, a piece of software called Hawkeye. The origins of Hawkeye are fairly simple — we recognised very early that a *proactive*, rather than *reactive* approach to support was critical in ensuring Hosting Confidence. To that end, we developed a custom server monitoring and notification system that alerts our technicians to *both* server/network failures and, by proprietary methods, a very broad range of conditions and issues that could indirectly or directly lead to server/network failures. It is the latter component that is, in some senses, the most important, because it is what enables us to very often stop problems on both our own servers, and our clients' servers, before they escalate into issues that lead to service degradation or downtime.