Martin Fenner interview

September 14th, 2009

The Future of Green in Storage

August 19th, 2009

It’s evident much is being done right now related to the “green” movement in computing, and there’s particularly a lot of such talk in the storage space.

Pranah’s CTO and cofounder, Martin Fenner, thinks one of the biggest areas to watch in this regard is the hard drive itself.  He thinks this is one area of “green” that can really make an impact. We sat down recently to discuss the topic of saving power.

“For instance,” he says, “let’s say you have a cooling fan on the system that uses — let’s say, just for example, 20 watts. We can try to make an effort to bring that down to 19 or 18. That’s fine, I’m not going to lose any sleep over that.  If we can reduce it to 18 with very little cost — sure, it’s nice to look into, but it isn’t what’s really going to affect the bottom line.”

Martin points out that most high-end hard drives, as of about 2008 and maybe early 2009, were in the 17-18 watt range.  That’s what was considered typical, between 17 to 20 watts, for a high-end drive.

Now, he points out, Seagate and other high-end manufacturers have a new generation of high-end drives that are in the 8-9 watt range. “Put that into the equation and it cuts electricity usage in half. That’s where it makes a difference. And, once you multiply that by 10,000, by one million, by 10 million, that’s when ‘green’ really starts to mean something,” Martin says.

The other thing that starts making sense, he says, is reducing power in any support electronics that go into the design of a system — especially of the smart variety, like controllers. “Any controller that has sizeable power, which they typically do in Pranah’s case — ours are about 85 watts — if you can find a way to bring that down at all, that’s real power savings.”

Martin continues: “So, let’s say you bring the savings from 85 watts down to 60. That may not seem much, but it is — it’s 25 watts.  That’s a light bulb.  If you can save that times thousands and thousands and thousands —  yes, that is significant savings.”

Our chat took us even further forward into the green movement, to the topic of solid-state drives. “The first wave of drives that came out were in 2-watt range for about an 80-gig drive,” Martin pointed out. “The only problem with that is the cost of the drive is still not where it needs to be. But that curve is obviously getting different everyday. We might get there one day — five, ten years out, we might be to a place where magnetic drives are history. That’s a possibility. We just don’t know that at this time.”

But Martin is not holding his breath for this day to arrive. “If you look at the cost and the curves and all that, it would indicate this is not going to happen anytime soon. But we never know. Look at the newer drives. Intel is basically the leader in solid-state drives. Their latest arrival of 160Gb drives is, I think, less than half a watt in usage. That’s when it starts making sense to talk green.”

What about the use of laptop drives as a potential for enterprises? I ask. “Well,  there are issues with that. There’s a place to use a solid-state drive in a laptop.  But there are places where using solid-state drives just plain does not make sense.  Obviously, all this will be controlled by costs. That’s where it will be determined,” Martin says.

“If a one-terabyte magnetic drive today were to cost the same as a solid-state drive, the race would be over. Okay? Real simple. But the race is not over — we’re not there. It’s still orders of magnitudes apart. There are drawbacks and advantages for both technologies that need to be very, very clearly analyzed, to see what you’re doing, because there are huge drawbacks with solid-state drives, and there are huge advantages with the magnetic platter. And let’s not forget: there are instances where a magnetic platter is faster than a solid-state drive. That’s a cold, hard fact.”

Again, said Martin, this may change over the years. “But just because solid-state drives are using less power and they’re available, it doesn’t mean that they’re going to rule the world and replace everything. That is not going to happen anytime soon.”

He went on: “One must realize that the lifespan of a solid-state drive is a serious issue as well. I mean, yes, they’re improving this stuff all the time. They’ve got new multilevel cells where you can write to cells multiple times. But still, in certain environments, solid-state drives have no place. It needs to be very clearly identified where they go.”

I asked Martin to consider products in the entry-level storage marketplace. What are some of those design considerations, from an engineering standpoint?

“Well, everything has to be designed with cost in mind,” he said. “Of course, the quickest way to design something is to go the X86 route. They’re cheap. There’s a lot of software out there for it. But it’s not the best architecture to do storage with. The X86 is a desktop CPU,  to run applications — Excel, Word, and what have you. These chips are not made to run storage applications. Storage applications don’t really care about compute cycles, necessarily. There’s another point that really applies to speed, and that’s I/O. X86 CPUs don’t have I/O, at least not enough, not for the kind of applications that we look at. Now, that doesn’t mean they’re not being used. Yes, they are being used. And I’m sure over time these CPUs are going to get faster. But there are better design philosophies out there than using X86 CPUs, and that’s where we’re focused for the future.”

Dave Walstad

VP Sales and Marketing

It’s evident much is being done right now related to the “green” movement in computing, and there’s particularly a lot of such talk in the storage space.
Pranah’s CTO and cofounder, Martin Fenner, thinks one of the biggest areas to watch in this regard is the hard drive itself.  He thinks this is one area of “green” that can really make an impact. We sat down recently to discuss the topic of saving power.
“For instance,” he says, “let’s say you have a cooling fan on the system that uses — let’s say, just for example, 20 watts. We can try to make an effort to bring that down to 19 or 18. That’s fine, I’m not going to lose any sleep over that.  If we can reduce it to 18 with very little cost — sure, it’s nice to look into, but it isn’t what’s really going to affect the bottom line.”
Martin points out that most high-end hard drives, as of about 2008 and maybe early 2009, were in the 17-18 watt range.  That’s what was considered typical, between 17 to 20 watts, for a high-end drive.
Now, he points out, Seagate and other high-end manufacturers have a new generation of high-end drives that are in the 8-9 watt range. “Put that into the equation and it cuts electricity usage in half. That’s where it makes a difference. And, once you multiply that by 10,000, by one million, by 10 million, that’s when ‘green’ really starts to mean something,” Martin says.
The other thing that starts making sense, he says, is reducing power in any support electronics that go into the design of a system — especially of the smart variety, like controllers. “Any controller that has sizeable power, which they typically do in Pranah’s case — ours are about 85 watts — if you can find a way to bring that down at all, that’s real power savings.”
Martin continues: “So, let’s say you bring the savings from 85 watts down to 60. That may not seem much, but it is — it’s 25 watts.  That’s a light bulb.  If you can save that times thousands and thousands and thousands —  yes, that is significant savings.”
Our chat took us even further forward into the green movement, to the topic of solid-state drives. “The first wave of drives that came out were in 2-watt range for about an 80-gig drive,” Martin pointed out. “The only problem with that is the cost of the drive is still not where it needs to be. But that curve is obviously getting different everyday. We might get there one day — five, ten years out, we might be to a place where magnetic drives are history. That’s a possibility. We just don’t know that at this time.”
But Martin is not holding his breath for this day to arrive. “If you look at the cost and the curves and all that, it would indicate this is not going to happen anytime soon. But we never know. Look at the newer drives. Intel is basically the leader in solid-state drives. Their latest arrival of 160Gb drives is, I think, less than half a watt in usage. That’s when it starts making sense to talk green.”
What about the use of laptop drives as a potential for enterprises? I ask. “Well,  there are issues with that. There’s a place to use a solid-state drive in a laptop.  But there are places where using solid-state drives just plain does not make sense.  Obviously, all this will be controlled by costs. That’s where it will be determined,” Martin says.
“If a one-terabyte magnetic drive today were to cost the same as a solid-state drive, the race would be over. Okay? Real simple. But the race is not over — we’re not there. It’s still orders of magnitudes apart. There are drawbacks and advantages for both technologies that need to be very, very clearly analyzed, to see what you’re doing, because there are huge drawbacks with solid-state drives, and there are huge advantages with the magnetic platter. And let’s not forget: there are instances where a magnetic platter is faster than a solid-state drive. That’s a cold, hard fact.”
Again, said Martin, this may change over the years. “But just because solid-state drives are using less power and they’re available, it doesn’t mean that they’re going to rule the world and replace everything. That is not going to happen anytime soon.”
He went on: “One must realize that the lifespan of a solid-state drive is a serious issue as well. I mean, yes, they’re improving this stuff all the time. They’ve got new multilevel cells where you can write to cells multiple times. But still, in certain environments, solid-state drives have no place. It needs to be very clearly identified where they go.”
I asked Martin to consider products in the entry-level storage marketplace. What are some of those design considerations, from an engineering standpoint?
“Well, everything has to be designed with cost in mind,” he said. “Of course, the quickest way to design something is to go the X86 route. They’re cheap. There’s a lot of software out there for it. But it’s not the best architecture to do storage with. The X86 is a desktop CPU,  to run applications — Excel, Word, and what have you. These chips are not made to run storage applications. Storage applications don’t really care about compute cycles, necessarily. There’s another point that really applies to speed, and that’s I/O. X86 CPUs don’t have I/O, at least not enough, not for the kind of applications that we look at. Now, that doesn’t mean they’re not being used. Yes, they are being used. And I’m sure over time these CPUs are going to get faster. But there are better design philosophies out there than using X86 CPUs, and that’s where we’re focused for the future.”
Dave Walstad
VP Sales and Marke

Pranah Storage System Design: Hardware Component Interchangeability and Other Design Considerations

June 21st, 2009

When we set out to design this 1U product, with so much more contained in one box than other systems, we wanted it to be a “tool-less” system for the customer.  I had worked with hardware systems for many, many years, and it was always a big frustration to replace parts and system components that were not hot-swappable or required various tools.  You always waste a lot of time trying to get parts out, getting fans replaced, getting disk plates replaced.

So, we came up with a unique design.  From the get-go, we wanted a system that, once it’s installed in the field, you don’t need any tools to get any parts out of the system.  Everything that can be replaced in the field — i.e., that’s meant to be replaced by users or service personnel — does not require tools.  That applies to drives, power supplies, fans, displays, adapters, you name it.  And the parts obviously that are not hot swappable are the ones that will be sent back as a complete unit, if that is ever necessary. For example, the motherboard itself is not hot swappable, because that’s a part we don’t want the customer to touch. It’s part of the unit itself.

Let’s move on to other design considerations, including the various host interfaces. Our premise for this whole product concept was that we wanted to come up with a so-called Swiss Army Knife.  You open the blade you need at the time.

There are really only four interfaces adopted today, and those are Fibre Channel and iSCSI, both 1 gig and 10 gig — of course, there are more coming in the future — and then the SAS/SATA standard.  Those are the only four that are really being used today.  We wanted to make sure we had them all in one system, and that they could communicate with each other.  So you can use any of these, and you can choose any route you want that’s valid.  You can go from Fibre to SAS/SATA drives.  You can go from iSCSI to Fibre drives or SAS/SATA drives.  It’s completely interchangeable.  We wanted to make sure that capability was being provided to the customer, and that is what we have achieved with the Pranah 2000 Series storage system.

Martin Fenner, CTO and Founder

A Look at the Hardware Design of the Pranah Storage System

June 19th, 2009

Let me describe the hardware design of the Pranah 1U in general terms, and how it’s different or unique. The design is based on an Intel XScale chipset.  What’s unique about this chipset is that, first of all, it has a dual-core architecture.  It’s ARM-based and it has dedicated XOR hardware built in, which helps us in generating the RAID functionality.  That’s basically at the heart of the system.

Then, the communication to the outside world is based on a PCI Express bus.  And we use an external switch to connect all the other devices to it, which are fibre channel, 10-gig Ethernet and 1-gig Ethernet.  There is an FPGA subsystem for the cache and it is used for inter-controller communication, so that we have a fast link between the two controllers. We can exchange information that way.

The Pranah system is designed such that it’s a control head that also offers drives.  It offers fibre-channel drives, SAS, and SATA.  All three drive types in one system — a single, one-use system.  The interface to the outside world has a fibre interface as both the target and initiator.  It has 10-gig and 1-gig iSCSI, obviously only as a target at this time.  And then it has SAS/SATA, as an initiator only at this time. (That might be a limitation on the Intel side, but that’s not a big limitation in SAS. That is not a network protocol. That might be remedied in future designs.)

That’s an overview of the electronics.  Now let’s talk about the mechanical design of the product.  The design of the Pranah 2000 Series storage products is based on an earlier JBOD design.  We designed the fibre-channel based JBOD first to kind of prove the high-speed technology.  There was a lot of physics involved in that — making it work in such a small form factor.  And so the first go-around was a 4-gig fibre-channel JBOD that holds eight drives in the 1U form factor, which brought some challenges in cooling, obviously.

So, we solved that then by coming up with a unique cooling mechanism, whereby we’re cooling from left to right in a rack, rather than front to back like everybody else.  That led us to design the special “open” side rails we now have in our system.  With these, we could basically blow out air on the right and suck in cold air on the left, thereby generating an under-pressure situation inside the box and providing superior cooling over the old front-to-back technique.  Eventually, this led to us apply for a patent, since nobody’s ever done this before.  So, we now have that patent on the design to cool sideways through the side rails.  That patent was issued in, I believe, May of 2005.

Let me explain a bit about the advantages of left-to-right cooling as opposed to front-to-back.  One of the big problems with front-to-back, especially in a 1U system — it may not be as clear in a larger-sized system — is that in front you typically have displays. You have buttons, you have all kinds of interferences.  And the same in the back: you’ve got connectors, you’ve got cables, you’ve got power supplies.  So, there’s really only a very small area of air flow that you can with front-to-back.

When you look at left to right — in our case, our system is 24 inches deep, which is about the max you should go to to be standard — the sides are completely unused.  There’s no obstructions anywhere.  So, cooling going from left to right is extremely efficient.  There’s no obstructions on that.  That’s what we’ve done –  about 80 percent of the area, left and right, is physically open for cooling.  Such a cooling system is not limited to only Pranah storage systems.  This patent could apply to other designs as well, not just storage.  So there’s a big upside for us with that patent.

Martin Fenner, CTO and Founder

Welcome to The Pranah Storage Blog

June 18th, 2009

Thanks for your interest in Pranah Storage Technologies, and welcome to our web site and blog.  We are excited to be announcing our company and product to the world on Tuesday, June 23 — so watch for that news release to be posted in the “Company” section of our web site.  The Pranah storage solution has been in development for more than two years and brings capabilities normally seen only in high-end system to the entry-level and small business price class.

The objectives of our blog are to communicate the latest developments about our company, our product, our team, our partners, and other timely news such as events we’re attending, customer success stories, software updates, and more. Others at Pranah besides myself who will be blogging will include our CTO and cofounder, Martin Fenner, and our VP of sales and marketing, Dave Walstad, and others ongoing.

Our hope with the blog is to engage with customers, prospects, media, analysts, other bloggers, and anyone interested in our storage technology. So, please comment on our blog posts and let us know your thoughts, questions, or concerns — or contact us by phone or email, as per the “Contact Us” page on our web site.

Sincerely,

Steve Carter, CEO