New Flash Entrants Validate QLogic Game Plan [View article]
QLogic is like an oil company that has reserves to last for several more years, but is drilling test wells for the future. Exploration is risky and expensive, but if you're an oil company, ya gotta do it.
New Flash Entrants Validate QLogic Game Plan [View article]
I peg the market opportunity for Mt. Rainier (PCIe Cache Adatpers for Enterprise applications) at somewhere from $200M to $500M in the next few years. More like $200M is the disk array vendors continue to influence IT organizations to add SSD to the disk array. More like $500M if IT migrates to putting their hot data on the PCIe Cache adapters inside the server for absolute best performance...like they should. Haven't seen any OEM announcements, but anxiously waiting for one like everyone else. 16Gb is not growth for QLogic. However, it does keep the Fibre Channel cash cow strong and healthy for several more years.
New Flash Entrants Validate QLogic Game Plan [View article]
Virtual IO systems from Xsigo, Virtensys and NextIO were invented in the 2005 time frame to provide a low cost alternative for connecting servers with high-speed, expensive 10Gb Ethernet. In 2008 HP was the first to bring out servers with 10GbE ports on the server....free. Since then, one-by-one, these companies have been acquired for their virtual IO intellectual property. The system architecture is dead.
Regarding your second point, today every PCIe SSD from, QLogic, Fusion-io, Micron, LSI, Virident, etc., which is used to cache a SAN disk array, requires a Fibre Channel HBA alongside the PCIe SSD. In the future, QLogic will merge two into one for a higher priced HBA. Overall, I think that is a good thing.
After years of winning wars of attrition in their respective markets, leading adapter vendors from Ethernet (BRCM and INTC), Fibre Channel (ELX and QLGC), and HPC (MLNX), have been thrown into the next round of the playoffs where all their features are combined into one class of product. So, it's not so much who has the most to gain--the question is "who is going to survive the next round?"
Where they're at today reminds me of the SUV market when it was new. They've made models that are wimpy trucks and uncomfortable cars. The eventual winner will figure out how to make models that are good at everything. However, today each company is missing pieces needed to break from the pack.
BRCM, INTC and MLNX don't have proven Fibre Channel drivers and installed bases of proven Fibre Channel products. If one of these companies combines with ELX or QLGC, they can breakout.
ELX and QLGC have done the best at creating products that are well rounded, but they are struggling to expand from their starting position of Fibre Channel (20% of network ports), and uprooting INTC and BRCM who dominate Ethernet (80% of the network ports in the data center). So long as BRCM and INTC do not Fibre Channel, they leave the door open for ELX and QLGC to get better at Ethernet and expand their share. If ELX and QLGC can introduce better 40GbE converged network adapters in the next year, it could be an inflection point for them. If ELX and QLGC can make 40GbE converged network adapters with hardware accelerated RDMA, they could also expand into HPC and impact MLNX.
MLNX has entered the HPC niche of the Ethernet market, and has no Fibre Channel products.
The industry has already decided the winning wire will be Ethernet, and Fibre Channel will be added to the long list of data, voice, and video protocols that are converged onto Ethernet. What is left for debate is the rate of the transition from native Fibre Channel to FCoE. For the first time, annual revenue for the latest generation of Fibre Channel (16Gb) is projected to be lower than the run-rates for previous generations. As 40Gb Ethernet (which brings with it 40Gb FCoE) emerges in the next year, FCoE will be even more competitive vs. 16Gb Fibre Channel, and FCoE growth will accelerate.
Perhaps. Together, they would have 90% of the Fibre Channel adapter market with an opportunity to consolidate development to improve profitability. My opinion is best matches for ELX are BRCM, LSI and MLNX.
FormerStorageGuy-- Do you believe Ethernet will transform itself to displace InfiniBand--similar to the way Ethernet adopted FCoE in order to annex Fibre Channel network traffic? Regards, Frank
If so, in the future will Fibre Channel HBA vendors need SSD functions to compete? And will PCIe SSD vendors need Fibre Channel HBA functions to compete?
Typically, a single Fusion-io card is installed into an application server to accelerate IO. Only one is installed because without fail-over capability, there is no need to install pairs. It also means you don't have the high-availability needed by business-critical applications.
If HA is needed, IT deploys a pair of external SAN arrays with fail-over--which is standard. To meet this requirement, Fusion-io stuffs a server of your choice with PCIe cards, which transforms them from compute servers, into external SSD SAN arrays. When deployed in pairs and connected together with a 40Gb inter-processor connection, they can be used in business-critical environments.
What QLogic does is allows the PCIe SSD cards to be spread across servers, so the capacity can be shared, and fail-over can occur. The inter-processor connection which maintains the heartbeat between QLogic SSD cards is the Fibre Channel SAN.
My opinion is all the PCIe SSD guys are gonna have to do this to compete in the future.
Signs Point to Dell Acquiring Brocade - What It Might Mean for Competitors [View article]
In my opinion, ELX 10 gigabit Ethernet chips represent the best converged networking chips for LAN-on-Motherboard available today. Designs wins at HP are proof. Also consider that Cisco makes their own server chips and switches which go inside their servers. This allows them to control development of unique network services between servers and switches. If HP and IBM believe in this model, ELX has valuable technology.
Brocade Racing to Eat Some of Cisco's Lunch [View article]
I agree Cisco has thousands of fabric customers. What Brocade and Juniper are others talking about is their new stuff that is shipping against Jawbreaker that is not shipping. The debate is, "Brocade and Juniper will translate faster time to market for this technology into market share gains," vs. "Cisco has been offering fabric technology and so much more to it's customers, that being a fast-follower with nex-gen fabric won't matter." - Love, Frank
Brocade Racing to Eat Some of Cisco's Lunch [View article]
As far as I know, there are no limit on chances. At every technology inflection point, the playing field gets leveled and the first-to-market tend to do pretty well. Innovate and they do even better. Brocade does not have an iPod here, but they are out first at an important inflection point.
Brocade Racing to Eat Some of Cisco's Lunch [View article]
Omar - I'll say it again, Cisco is the market share leader for Fabric solutions. Regarding FCoE, there is a perception that Cisco is pushing convergence on FCoE, while the hot technology with customers is next-gen (Jawbreaker) flat networks. Maybe the perception of Cisco pushing too hard on FCoE, and falling behind in Fabric technology was created by Cisco competitors. But the perception exists, and Cisco competitors are pouring fuel on the fire. This will be a tough marketing problem for Cisco for awhile.
Brocade Racing to Eat Some of Cisco's Lunch [View article]
Long Haul - Cisco is the leader overall in this space. But Juniper and Brocade are defining the market for next gen technology. I believe this article from Jim Duffy at Networkworld captures where Cisco technology is positioned in this space: www.networkworld.com/n...
I think Brocade is attractive to a big player that needs the networking piece of a complete cloud stack. To get their portfolio, that player is willing to sacrifice the loss of current OEM customers they will compete against.
New Flash Entrants Validate QLogic Game Plan [View article]
New Flash Entrants Validate QLogic Game Plan [View article]
Haven't seen any OEM announcements, but anxiously waiting for one like everyone else.
16Gb is not growth for QLogic. However, it does keep the Fibre Channel cash cow strong and healthy for several more years.
New Flash Entrants Validate QLogic Game Plan [View article]
Regarding your second point, today every PCIe SSD from, QLogic, Fusion-io, Micron, LSI, Virident, etc., which is used to cache a SAN disk array, requires a Fibre Channel HBA alongside the PCIe SSD. In the future, QLogic will merge two into one for a higher priced HBA. Overall, I think that is a good thing.
Mellanox: Between Perfect Storms [View article]
Where they're at today reminds me of the SUV market when it was new. They've made models that are wimpy trucks and uncomfortable cars. The eventual winner will figure out how to make models that are good at everything. However, today each company is missing pieces needed to break from the pack.
BRCM, INTC and MLNX don't have proven Fibre Channel drivers and installed bases of proven Fibre Channel products. If one of these companies combines with ELX or QLGC, they can breakout.
ELX and QLGC have done the best at creating products that are well rounded, but they are struggling to expand from their starting position of Fibre Channel (20% of network ports), and uprooting INTC and BRCM who dominate Ethernet (80% of the network ports in the data center). So long as BRCM and INTC do not Fibre Channel, they leave the door open for ELX and QLGC to get better at Ethernet and expand their share. If ELX and QLGC can introduce better 40GbE converged network adapters in the next year, it could be an inflection point for them. If ELX and QLGC can make 40GbE converged network adapters with hardware accelerated RDMA, they could also expand into HPC and impact MLNX.
MLNX has entered the HPC niche of the Ethernet market, and has no Fibre Channel products.
Mellanox: Between Perfect Storms [View article]
Mellanox: Between Perfect Storms [View article]
Mellanox: Between Perfect Storms [View article]
Mellanox: Between Perfect Storms [View article]
QLogic Growth Engine: SSD [View article]
QLogic Growth Engine: SSD [View article]
If HA is needed, IT deploys a pair of external SAN arrays with fail-over--which is standard. To meet this requirement, Fusion-io stuffs a server of your choice with PCIe cards, which transforms them from compute servers, into external SSD SAN arrays. When deployed in pairs and connected together with a 40Gb inter-processor connection, they can be used in business-critical environments.
What QLogic does is allows the PCIe SSD cards to be spread across servers, so the capacity can be shared, and fail-over can occur. The inter-processor connection which maintains the heartbeat between QLogic SSD cards is the Fibre Channel SAN.
My opinion is all the PCIe SSD guys are gonna have to do this to compete in the future.
Hope that helps.
Signs Point to Dell Acquiring Brocade - What It Might Mean for Competitors [View article]
Brocade Racing to Eat Some of Cisco's Lunch [View article]
Brocade Racing to Eat Some of Cisco's Lunch [View article]
Brocade Racing to Eat Some of Cisco's Lunch [View article]
Brocade Racing to Eat Some of Cisco's Lunch [View article]
I think Brocade is attractive to a big player that needs the networking piece of a complete cloud stack. To get their portfolio, that player is willing to sacrifice the loss of current OEM customers they will compete against.