Why Don’t Virtual Servers Get Calculated in Data Center Market Share?
Kent Erickson from data center monitoring service Zenoss has queried the accuracy of analyst reporting on server market share, suggesting the need for the market to shift to a model that recognizes the growing hybrid cloud infrastructure alongside bare metal players.
Riffing off the latest Gartner analysis, Erickson queries why server market share is still calculated as a split between those who ship bare metal servers without taking large cloud data centers, like Amazon, into account.
Other analyst shops like IDC also calculate server market share as being a race between the same five bare metal providers as Gartner: HP, Dell, IBM, Lenovo and Cisco (with everyone else lumped together as “Others”).
Erickson argues that the time is right to include virtual data centers in the market share mix, pointing to VMware’s and Amazon Web Services’ (AWS) servers in particular.
Melding first quarter data from Gartner with his own research on second quarter financials data from AWS and VMware means Erickson has created a back-of-the-napkin estimate that purists could rip up. I’m not sure why for his own arguments sake he didn’t use first quarter data to maintain consistency. But in any case, the picture alters significantly enough to suggest that Gartner and other analysts are missing a crucial part of the data center overall picture, especially in a year when hybrid cloud is becoming a growing decision amongst the enterprise.
Erickson estimates that amongst the virtual and bare metal pie, VMware and AWS account for some 20 percent of all servers purchased, with Amazon’s 11 percent being second highest of any player, with only HP’s 19 percent outranking them. What’s more, Erickson argues that Amazon’s market share has doubled in the past year, making them on track to dominate the market in only twelve months.
Calling out latency, reliability and security as unanswered indicators in the measurement of market share, Erickson poses the question: “Are you ready to have the majority of your new servers running outside the walls of your data center?”
While creating some inconsistencies in using differently sourced calculations to create an overall industry view, the data is compelling enough to suggest that a new model of measuring hybrid cloud infrastructure uptake may be needed to better reflect current enterprise spending on data infrastructure.
Cisco, HP and IBM are sponsor of The New Stack.