by ramdas published July 29, 2018, 10 a.m.
There is a gross lack of innovation and imagination in the x86 server marketplace!
That statement might come across as strong. However outside of the enterprise data center marketplace which every vendor is eyeing, there's nothing really happening.
And if you are a product manager at Dell, HP or Lenovo you will soon smell the coffee. Enterprise market will shrink, or they will start moving to the public cloud. They have been reluctant so far, mostly because CIOs are actually fuddy old control freaks. But soon public cloud will emerge as an attractive option to the CFO. The calculators will be out and you have smart sales guys from Amazon, Google and Microsoft who will drive the public cloud adoption. Soon consultants will start advocating the public cloud advantage.
And none of the cloud vendors will buy hardware from branded vendors in the future. They will shop in China, or build it themselves. If you really want to be competitive as a public or private cloud vendor then you need to squeeze every bit of your bill of material costs to stay afloat. When you have a million servers in your farm, a $ 10 advantage is $ 10 million in the bank.
My sources at Flipkart says that they are abandoning Dell to buy server hardware from some Chinese vendor. I have not got an official confirmation, but I won't be surprised if it is true.
Google and Facebook have been building hardware for ages. Facebook's Open Compute Project has been getting decent press, though I really don't see it going beyond a few dozen large customers. Of course, there exists a brilliant opportunity to build hardware around OCP and sell it to SME and even mid-market customers. But this is a game for Gigabytes and Tyans of the world.
I would bet that by 2017 one of out of two servers running will be in a data center belonging to large internet player such as Google, Facebook, Microsoft or an Amazon. And there would be around a dozen such players who would be building servers either themselves or through contract manufacturers. They will support these machines themselves, essentially throwing a hardware out in less than 15 minutes, and replacing it. None of the current service level contracts offered by branded server vendors can match this.
A number of customers across all sectors from small business to the largest enterprise will shift base to a public cloud.
Not all of them will move, but some of them will move. And those who move, may not move everything. That's why hybrid cloud concept is so exciting.
However, implications are very clear on server hardware front. This market will shrink for large vendors, unless they innovate.
For the past five years or so, the commercial server market has become a commodity bazaar. Problem with commodity market is the brand gets diluted faster than you think, and without lack of innovation you really cannot differentiate.
Look at the stock and sell market for servers. Each of the three top vendors carried between 4 and 12 models. Dell had 4, while IBM/Lenovo and HP had a dozen each. Take the entry-level servers of all the three. HP and Dell had exactly the same specification. A single Gigabit Ethernet port Server featuring Intel Xeon E3-1220 Server. How do you create a differentiation when you are are building products that look like one another. How can a Dell confidently prove to a customer that their box is better than that of HP?
A few months back, I had bumped into a senior product manager at Dell and shared my observations. He said his hands are tied, and admitted that most of the product strategy comes from the headquarters. He also admitted that all these years Dell simply competed with HP and was focused on building models that matched spec on spec, but with Dell's legendary efficient built-to-order model was cheaper to manufacture and cost less.
However, Dell is today emerging into a distribution and channel-driven company and their logistics are not quite different from the competition these days..
When I heard for the first time about IBM's plans to sell to Lenovo in mid 2013 (the actual announcement happened in early 2014), my initial reaction was a disappointment. IBM is a smart company, and has at least in last two decades made moves ahead of the market. It got out of peripherals business 20 years back, got out of home PC business 15 years back, commercial PC business 10 years back. And we know what happened in each of these segments.
It's clear that IBM saw no future in the x86 server business. For Lenovo which is the PC market leader globally and makes about 1-2% after PAT in the business will know precisely how to run a commodity business.
However, at least at this point of time, I am yet to see any innovation from Lenovo on its own. The System X brand and the technologies it brought along will do well for Lenovo in gaining a decent foothold in the server market.
Outside of the enterprise data centre and cloud vendor market, there exist the enterprise non-data centre, mid-market and SME space. These are markets which also can go the cloud way, but with innovation, vendors can still hope to hold on.
How can server vendors be disruptive in an increasingly commoditized marketplace? I will discuss this in the next post.