Watch out blade server market – here comes ATCA!
This past week I attended the Light Reading ATCA & Communications Ecosystem Conference in San Jose, CA. This live event attracted about 120 people including ATCA system developers, ecosystem vendors, and consortia. Here is a quick list of the companies I recognized at the event:
- TEMs/NEPs – Ericsson, Nokia-Siemens, Nortel, Stoke Networks, Cisco
- Vendors – Continuous Computing, Emerson Network Power, Enea, GoAhead, Intel, Kontron, RadiSys, Wind River
- Consortia members – CP-TA, Linux Foundation, PICMG, OpenSAF, SAForum, SCOPE Alliance
(check out Glenn Seiler's blog on the Telecom Consortia Alphabet Soup)
The event was somewhat low-key, probably a reflection of the economic times. However, the content was great. ATCA has been deploying for a few years now. The market has been picking up steam as more networks move to an all IP infrastructure. We can thank the acceleration of smartphones and mobile internet devices (MID) for fueling the demand. And, by the looks of it, this market will continue to grow as equipment providers begin to design in Long Term Evolution (LTE) technologies.
There were a few great data points, especially related to the Server market. Here are my notes from the event:
- The ATCA market is projected to reach $6.8b by 2012.CAGR at 93% (courtesy of Earlswood Marketing, April 2009)
- Growth for ATCA-based equipment will require penetration into new areas of the network – perhaps the bladed server for network datacenters?! There were a few Cisco folks in the audience, any correlation?
- The 3G data plane market is exploding with the control plane remaining relatively fixed
- COTS system solutions are gaining ground on in-house designs. It is estimated the split will be 50/50 by 2010. A few years ago we estimated 90% in-house, 10% COTS.
- The second generation of ATCA-based equipment is rolling out now. These are 10Gb systems with dual and quad core CPUs. They are used in wireline and wireless 3G/4G infrastructure equipment.
- The third generation of 40Gb/s ATCA-based equipment is in development. It will include CPUs with 8 or more cores
- Interoperability remains a challenge
- More needs to be done to further open architectures and open interfaces. This includes Carrier Grade Linux OSes and HA middleware
- Virtualization is important but there are challenges that still need to be overcome. Today the different components that make up an ATCA system are not ready for virtualization. ie: HPI
- Linux is the least common denominator operating system – Carrier Grade Linux is absolutely necessary to reach the 5-9s availability – there cannot be any weak links
- Hypervisors will become even more important, but they must be Carrier Grade
- A new opportunity to provide system integration to Service Providers is emerging. This will center around interoperability and integration services
Now that I did a huge data-dump, what are your thoughts? What does Cisco's entry into the server market mean for IBM Blade Center, HP and NOW Oracle/SUN? Why is Oracle entering the Server hardware market by acquiring SUN? Did IBM miss a big opportunity to keep Oracle out? Who will use ATCA as a competitive advantage? Shouldn't Carrier Grade performance be standard in Data Center servers to drastically reduce total cost of ownership and realize even greater availability? Check out this white paper on Carrier Grade Linux: High Service Availability for the Next-Generation Network (registration required).
Want to learn more about the ATCA market? Join Wind River at the MicroTCA Summit.