Something’s gotta give

Posted on April 3rd, 2010 by Cheyne | Linked in Opinion

If you’ve recently been wondering about the state of the Virtual Private Server (VPS) segment of the Australian web hosting industry, then don’t worry, you’re not alone.

We’ve watched with amusement as more and more “budget” VPS services flood the market, offering everything for next to nothing, while at the same time watching customers come to us on Live Chat or eTicket asking when our Xen VPS services will be available again, as they’re annoyed with the speed of the service, constant high loads, disk I/O problems, or just issues with customer service in general.

The idea behind server virtualisation is actually a good one, but it was never designed to be an oversold platform for cheap containers. In fact, it was designed to make use of under-utilised servers where CPU time and memory otherwise went begging, thereby getting more “bang for your buck” from one server.

The problem with cheap VPS’s is the price. As we found with our Deal of the Decade shared web hosting plans, price is generally a deal maker, but when you’re talking about cheap VPS’s you need to be able to cram in dozens of containers on to the one server (or blade) to make it even come close to turning a profit, and therein lies the problem. A virtual private server should be exactly that, a private server where it looks and feels like nobody else exists, but when you’re cramming 30+ VPS’s on to the one hardware node, you’re going to have issues… eventually.

There are two good examples in the Australian market that I would draw your attention to.

The first is that of Crucial Paradigm. Beside the fact that they still advertise their VPS offering as a “cloud” service (when they tell you it’s not technically a cloud service, but just a marketing slogan), but during a recent hardware failure on one of their VPS nodes, they said they have around 35 VPS’s on each hardware node.

This means at any given time, up to 35 VPS’s can be competing for CPU time, memory and disk I/O, and network access on the one shared ethernet port. On top of this, they have persisted in using HyperVM, a platform for managing virtual servers, even after it was exploited in June 2009 which resulted in over 100,000 websites being wiped out and the owner and founder of the company tragically committing suicide, leaving the project in dire straits.

The second example is Jumba. When you choose to run a virtual server container (VMware), with a virtual server platform (Virtuozzo) to provision VPS’s, automatically the alarm bells ring. Virtuozzo may be the “best value virtualisation platform”, but it’s also one of the most clunkiest and restrictive platforms available.

The trouble is that it’s not a true hardware-based virtual platform, it’s a full software virtual platform that simply emulates hardware, which requires special kernels made by Parallels in order to operate correctly. This causes problems for many users, as the kernels are often many versions behind the “current” release versions of kernels which may be a security risk, in addition to many core features that are missing and unable to be included (until recently, iptables was extremely difficult to operate in a Virtuozzo container).

They are also spruiking the benefits of their IBM blade hardware and 48GB of DDR2-533 memory. Ignoring the fact that DDR2-533 memory is quite old in the grand scheme of technology in 2010, knowing the operators as well as I do, you can be assured that every megabyte of that 48GB will be used.

Going by what most people are saying on their forums and on Whirlpool, most are choosing the “Vegetarian” option which gives them 512MB of RAM, meaning they plan to squeeze around 90+ VPS’s on the one blade! Even if we discount down to 60 VPS’s on the one blade (to allow for a couple of the other menu items), that becomes 60 VPS’s all clawing for CPU time and memory I/O, not to mention the fibre channel disk I/O – but remember, this is just one blade of up to fourteen in the one chassis, all sharing the same fibre channel disk I/O. And because the disk space allocations are so low, the fact that the SAN is running 15,000RPM SAS drives becomes completely irrelevant, as one hard drive could be seeking data for up to 20 VPS’s at any one time.

And on top of this, we haven’t even began to touch on the economics of either example. Our first example would be outlaying around $10,000 for each of their nodes based on the advertised configuration, for a $1,400 per month return, not taking in to account network, co-location and support costs. While our second example has shelled out around $150,000 for their hardware, which is why you can expect them to cram as many customers per blade as possible, as even 60 customers on the base plan only brings $597 per month in to the till, which wouldn’t even cover the cost of it’s share of the monthly repayment, let alone network, co-location or support costs.

While I sincerely wish both of these companies all the best in their endeavors, I will admit to being concerned about how customers see this type of offering without fully understanding the impact it could potentially have on both their service, and the market in general.

The great thing for us is that no sooner do we provision new VPS hardware nodes they are full, and with a waiting list of more than 40 customers, we know that our service, a true hardware-based Xen virtual server environment that doesn’t oversell CPU time or memory I/O, is still a drawcard for us.

We just need to make sure customers are better aware of the different types and levels of service available, and that when they are looking for a VPS solution, they compare apples with apples, and not lemons.

I accept cash or card, but no cheques, thanks.

Posted on February 15th, 2010 by Cheyne | Linked in Opinion

It’s always amusing when I see our creativity being utilised by other companies, and however flattering it may be, you would at least think that companies who are much larger than us employ capable and competent business analysts to set the standard, rather than just keeping up with the Jones’.

With the launch of our new web site last week, we re-introduced the “Live Sales Chat” function after the majority of Australian web hosts had switched the service off. For our business model, it’s a great benefit for new and existing customers,  ensuring they are able to get the answer to sales-related questions quickly and easily.

Today, a competitor has launched a “Live Sales Chat” service on their web site, presumably after seeing it on our new web site. They’re more than welcome to the idea, in fact, I won’t be at all surprised if I see giant LCD screens or an integrated FAQ system on their site sometime soon.

This isn’t the first time this has happened to us. Our marketing strategies and offers are often replicated by competitors who simply lack the capacity to think outside the square.

While it’s humbling to know that we’re truly leading the way in terms of innovative ideas and automative productivity, I can’t help but feel like I’m a pseudo consultant for these companies. My usual rate for consultancy work is around $200.00 an hour, however I’m willing to apply our “50NOW” coupon taking 50% off the first invoice.

I accept cash, credit card, but no cheques, thanks.

1, 2, 3, 4, who’s gonna start a price war? Not us.

Posted on February 8th, 2010 by Cheyne | Linked in Opinion

The chattering on Whirlpool in recent hours has made me smile, with the suggestion that another web hosts’ ingenious plan to “slash prices” would somehow start a price war with us. Boy were they mistaken.

What many people fail to understand is that we don’t make a decision without it first being discussed among our Management team, which consists of four people who have more combined Management experience in this industry then almost any other Australian web host, nor do we simply “slash prices” to compete with a company who has less than 200 customers in total.

If they’ve made the choice to “smash and grab” for new customers then good on them, but that’s not a strategy we employ.

Prices on our Economy cPanel web hosting plans is based on having 98% business automation through the VIPControl system, running our own network, customer feedback which highlighted the fact that some customers don’t need phone support and other services (but were willing to pay additional for it if and when they needed it), and here comes that word again.. experience.

Last week we released a special offer called the “Deal of the Decade”, which gave customers a cPanel web hosting account at just $39.00 for 2 years. Crazy? Maybe, but as promised, I’m going to explain my methodology behind this plan.

When we were discussing how the web hosting industry is very similar to the airline industry, I came to the realisation that a segment of the airline industry was not yet represented in web hosting form. You see, there are the “Qantas” web hosts like NetRegistry and MelbourneIT, who offer a business grade service with all the trimmings, and then there are the “Virgin Blue” web hosts like SmartyHost and Hostess, who offer a consumer grade service with optional extras and addons. But where are the Tiger’s and AirAsia’s of the web hosting industry?

Ultra low cost carriers exist for the sole purpose of generating revenue for their parent company, and for increasing buying power when negotiating the purchase of new aircraft with suppliers, which has an overall positive effect on the business, providing they continue to get bums on seats.

And they do (get bums on seats that is), simply because of the price. How many times have you considered flying away in recent times simply because you’ve been able to access $29 airfares?

So the question I asked was: If we offered $39 web hosting accounts with 2 years of service, would people buy it simply because it was a bargain? The answer is a resounding “yes”. In just one week, we sold all 500 plans at $39, netting us $19,500 in additional revenue. You may be forgiven for thinking that a plan like this is unsustainable, and to most cPanel web hosts it would be, but when our customers have access to VIPControl, the “cost” of supporting these additional customers shrinks to almost nothing.

Think about it a different way. Imagine you have 500 customers who all want to enable SSH access, fix the permissions of their account, and enable temporary URL access. To every other cPanel web host in the world, they would need to call up or submit an email/ticket to the support desk in order to have a persona manually activate the services on their account. Our customers simply login to VIPControl, choose the options and enable or disable them.

So, we apply the same methodology as the ULCC’s and ask ourselves, if this offer was available every month, would we be able to sell 500 plans (one full 1RU server) each month for 24 months, giving us a total increase of 12,000 customers (after 24 months) and an additional $19,500 of ongoing monthly revenue?

Well, we want to find out.

Later today, we will be re-releasing our “Deal of the Decade” plans with another 500 plans available, to see if the initial reaction was simply a one-off, or whether there truly is an untapped market waiting to be explored. If there is, we’ll consider releasing the service as a permanent item under a separate brand name.

The moral of the story is that competing on price is one thing, but being able to back up those prices with a quality product and market-leading customer service is another. We’ve proven our model works with more than 18 months of constant growth behind us, and our customers have never been happier with the service we deliver, both through VIPControl and our friendly and experienced staff.

Can your web host do that?