Tags


Latest Posts


Latest Comments


Authors

Safety in numbers

jonathon-birch-thumbnail.jpg

Posted by |

For IT teams needing to wrestle time and budget back from those housekeeping chores of hardware maintenance and server management, shared hosting has been a godsend. Indeed, it’s a perfect option for customers starting out on a managed cloud  journey: in having a share of an existing platform, your server is basically one of many customer servers running on a common piece of physical hardware, all utilising the same underlying resources such as CPU, memory, and storage. Shared resources also equate to shared costs but the benefits extend beyond pure savings. No capex requirement to delay deployment, dynamic and elastic scaling to cater for all future requirements, sound performance, paying only for what you use, and full control of your virtual machine (VM) server instance, these have all contributed to shared hosting’s simple recipe for success.

But some have remained hesitant, equating ‘shared’ to ‘not as good as’. In the abstract, you could argue that that is true and shared can’t compare with, say, dedicated hosting, which must automatically be better. But we’re in the real world and ‘good’ is judged in terms of how well the solution fits the purchaser’s needs at a time when increasingly it’s about balancing operational, financial and/or technical requirements with cost and feasibility. So ‘shared’ can’t be inferior if what matters to you is having cost certainty, affordability, freedom to expand, resilience, redundancy and availability and shared hosting delivers that day in, day out. In that situation, given those competing needs, ‘shared’ is indeed the optimum solution.

It would also be harsh to judge it as inferior where your provider has taken particular steps to address the resource allocation issue. If your server is hosted with a provider who is allowing clients to utilise a high percentage of underlying resources, then this may negatively impact the performance of your server. While some hosting companies may alert their customers who are using high resources on a shared platform through the use of monitoring agents and capacity planning tools, others, Redcentric included, prefer to exert more control over this situation from the outset. In our case, we apply a contention rate of 30% on our physical CPU use which allows for fair and effective resource allocation to existing and new tenants.

The earworm of ‘not as good as’ is perhaps most likely to sound loudly and persistently when it comes to one of hosting’s most fundamental issues – security. The tune will go something along the lines of ‘In a shared environment there is surely more of a risk of cross contamination between VMs, of customer A being able to see customer B’s data?’ It is most certainly a question that has been asked of hosting providers. It is also a patent nonsense.

It’s like saying a block of flats is inherently less secure than a house. Rather, it depends on the quality of the security provision in that particular instance. A block of flats - which, incidentally, is a good analogy for a shared server platform, with every tenant using the common resources like the car park, reception and lifts and yet maintaining the privacy of their own apartment - a block that is properly protected might boast gated parking, key code front door entry, card-operated lift, CCTV, security guards, concierge or caretaker and that's before you get to your own front door, which has a spy hole, locks and bolts, and an alarm. A house, on the other hand, could be locked with one single key - so not secure at all really.

Given that, it makes sense to retune that earworm, and have it as a constant reminder to check the level of security in place - whether you're utilising shared or dedicated hosting. With Redcentric, whichever you choose, you get the same gold standard of security. There are virtual servers connected by virtual network cards to virtual networks and these virtual networks are protected by virtual firewalls. These virtual firewalls might be virtualised instances of traditional physical network firewalls, or they may be special instantiations of network traffic control mechanisms that are integrated with the hypervisor environment. What they are not is inferior in any way – just like the shared hosting they secure.

Comments

 

Post a comment

Comment submitted! Comments needs approval before being displayed.