SV Adventure

The Power of Commodity Hardware

2009-01-20T15:01:13-08:00

I have learned a great lesson I wanted to share:

Commodity hardware trumps proprietary hardware.

Last Fall, DataScaler shifted to delivering our software on a cluster of commodity servers. Previously, we had planned to deliver our software on purpose-built custom hardware. The move to commodity servers has been one of the best decisions we've made as a company. Why all the love for commodity hardware?

Customers love commodity hardware. I was surprised how strongly our customers preferred the commodity servers option. When presented with both options, customers overwhelming chose the commodity hardware, even though our appliances were very elegant, compact, etc. Some customers are HP shops. Others are IBM shops. Some are Dell shops. Their IT teams know how to purchase, configure, rack, and manage their standard servers. Bringing a piece of foreign hardware into their environments is only done grudgingly.

Competing with HP, IBM, and Dell is dumb. With the major server manufacturers spending billions on server R&D, a small startup cannot hope to keep up. Our (very cool) appliance design packed a lot of compute power into a small space and did it with out melting. The appliance would have given us an 18-24 month lead over commodity hardware in terms of compute density. However, that lead is too short to recoup the costs involved. More importantly, commodity hardware will continue to get better every year, whereas our design would be stagnant. We could have invested many millions each year to update the hardware, but why? The incremental gains would be small and it's nearly impossible to keep up with the giant server makers.

Commodity hardware is much more capital efficient. By switching to software running on a cluster of commodity servers, we substantially reduced the overall capital needs of the company. Most importantly, we can get to market on less capital, allowing us to raise less money and give up less equity in the early phases.

Luckily for us, there was very little effort involved in porting our software from our purpose-built hardware to commodity servers. All our intellectual property is in our software, and the hardware, though cool, was not core to our business. Shedding our custom hardware has been a great decision, and I only wish we had done it earlier.

As they say, hindsight is 20/20.

Local Outsourcing

2008-03-05T22:47:17-08:00

In the last decade, we've seen outsourcing skyrocket, then drop off somewhat. Offshoring work to India has become so expensive and difficult that many companies have abandoned the practice altogether. Additionally, offshoring brings complications of communication, retention, management, etc. However, let's not confuse outsourcing and offshoring. While we are not proponents of offshoring in general, we find great benefit in outsourcing to local firms.

DataScaler is a hardware company that, under the covers, is really a software company. We build an appliance which installs in a data center and fulfills a particular role. The appliance is a very complex system and is not just a PC motherboard in a box with a logo on it. There are FPGAs, ASICS, and lots of custom circuitry that make the box tick.

When we founded the company, the core concepts were a business problem that needed to be solved and an overall architecture to solve that problem. The architecture requires custom hardware and software. That is a lot to bite off for a startup, so we partitioned the problem into what was core / differentiating technology and what was non-core. The core items are:

Our overall architecture
Our processing engine (implemented in an FPGA)
Our software.

Hardware products involve a bunch of things outside of this core, such as:

Chassis / mechanical design
Industrial design (look & feel)
Board-level design
Cooling design
Manufacturing & assembly.

Each of these areas is a specialty and requires experts. If we did these tasks ourselves, it would mean hiring experts, giving them stock, managing them, giving them offices and computers, etc. The cost and effort add up quickly. Also, finding experts in each of these fields who are ready to jump into a startup takes a lot of time and may not even be possible sometimes.

Thus, we decided to outsource the non-core aspects of our product. We do the system architecture, FPGA design, and software in-house. We outsource the rest. This has been one of the best decisions we've made at DataScaler. Prabhat, our SVP of Engineering and Operations, deserves the credit for this one. By outsourcing to local firms, we've gotten top-notch engineers who could start immediately on the project. The results have been great, and we have saved both money and time.

I also think there is value in having engineers who work on lots of other companies' products. This diversity of experience fosters new ideas and a continual exposure to different ways of doing things. Another advantage is that hardware projects are bursty: they require a lot of people at once, then go into maintenance mode for a while. This burstiness is difficult to handle when your engineers are all on your payroll. By outsourcing, we can ramp staff up and down as needed during the lifecycle of a particular product.

The trick in all this is choosing the right outsourcing partners. Again, Prabhat deserves the credit here, but I couldn't be more happy with the set of partners we have at DataScaler. Prabhat has been in the business for 25 years and has a very broad network. By tapping that network for referrals and vigorously vetting the candidates, we ended up with the dream team of hardware design:

Arira Design in Sunnyvale does the electronic part of our hardware design. This includes everything from translating our high-level architecture into schematics and PCB layouts, to component selection, to bringing up and debugging the first systems. Arira has done lots of complex products like ours and got great reviews from Woven Systems, who used Arira to design their 10G switching platform. Arira also acts as the project manager, coordinating the activities of the other partners. We couldn't be more pleased with the results.
Studio Red in Menlo Park does the mechanical and industrial design. I had worked with Studio Red in a prior company and was impressed with their skill and strong design sense. Having been at this for 25 years, Philip Bourgeois and his firm have done thousands of designs. That experience comes in handy when time to market and execution are critical.
Electronic Cooling Solutions handles the airflow and thermal issues of the design. These are difficult problems that often get neglected until a design is done. Having specialized expertise involved at the outset saves lots of money in redesign costs, etc. Bill Maltz and his team have consulted for the who's who of Silicon Valley and work very quickly. I admit that although I took a basic thermodynamics course in college, I am pretty clueless when it comes to how to manage the airflow and cooling of a complex, high power density appliance. Having ECS involved allows me to sleep well at night knowing there won't be thermal issues.
Mike King is a guru of the black magic involved in handling all the Electro-Magnetic Interference (EMI) issues inherent in a system of our type. With a plethora of 3-10 GHz signals running all over inside our box, the EMI challenges are difficult, to say the least. Mike has been doing this work for over 30 years and is the best there is.

To any company contemplating a hardware design, I highly recommend taking this outsourcing approach. It enables our company to focus on our core and differentiating technology while taking advantage of the deep experience of our partners.

Productivity

2007-12-28T09:30:45-08:00

During the time between Christmas Eve and New Years Day, we close the DataScaler office. This gives employees the chance to recharge, travel, be with family, etc. For me, it's been a very productive time. Without the distractions of meetings, conference calls, etc., I have gotten a lot done. I enjoyed some great family time as well.

Working at home is both better and worse for me than working at the office. At the office, I can focus solely on work, without my daughters' occasional interruptions. However, at the office I have nearly-constant interruptions from work-related things. Thus, when I have reading to do or some large project that takes solo time, I prefer to work from home.

The office is great for interaction, collaboration, and generally staying on top of the buzz of the company. We value having a physical office where everyone comes in every day and we work together. Although working from home may be slightly more productive, in a startup, things change rapidly, so communication is more important than raw productivity. Making great progress against the wrong goal isn't productive. Thus, we value being together, sharing ideas, interrupting each other, overhearing each other, etc. When I have a great customer call or new concept to bat around, my team is right there with me. We would communicate much less if we were not physically together.

We also have chosen an open office layout. We have no offices or cubicles. This is inline with our core value of open communication. When we have personal phone calls, etc., we make use of the conference room or take a walk outside. This arrangement works very well for keeping everyone on the same page as to what is going on. Again, it is not optimized for raw, uninterrupted productivity, but that is not our main focus. The communication and collaboration benefits far outweigh the cost of interruptions, etc.

With all that said, it is nice to get away from the office and get some heads down time. I've enjoyed the break so far and look forward to the next few days. Come Jan 2, when we all get back together at the office, I am sure we'll all have lots to talk about.

Our First Award

2007-12-14T21:01:32-08:00

We're very excited to announce that DataScaler was just voted the "Most Innovative New Technology of 2007" by 40 IT executives in the financial vertical. I just returned from Manhattan where I attended the annual event, which was organized by Cresting Wave, Inc. Ten technology companies each gave a short presentation. Afterwards, the 40+ IT executives (CIOs, CTOs, and VPs for major financial institutions) voted for the company they were most interested in. We're very honored to receive this coveted award and are even more excited by the interest in DataScaler which it represents. The attendees of this event included Bear Stearns, Deutsche Bank, UBS, Prudential, etc. These firms recognize the value that DataScaler brings and are chomping at the bit to get it. One of our big issues now is holding them off while we complete the product. ;-)

I thought you'd enjoy hearing this good news.

Happy Holidays,

Chad

Thursday Night Magic

2007-09-03T17:19:15-07:00

At DataScaler, we have a work-late party every Thursday night. We bring in dinner, invite our technical advisors, and great things happen. It's a magical time. We also have a great set of business/strategy advisors; I'll post about them another time.

We have excellent technical advisors. They come from various backgrounds, all in some way related to our technology direction. Some are more talkative, some are more contemplative, but they all contribute in great ways. I am amazed at the good things that come out of our Thursday meetings. During the week we work hard on the product line, its architecture, strategy, etc. When we present the current state of affairs on Thursday nights, often one advisor or another will point out a flaw we hadn't thought of, or a simpler way to do things. It's very helpful to have this group's attention once a week and benefit from their broad experiences.

It's invigorating to see how people jump in and worked hard as DataScaler technical advisors. We don't offer much in the way of compensation: free food and a token stock grant. However, everyone seems to get really excited about helping out with the technology. I can't say I blame them ;-)

Again, I am reminded how blessed we are to be in Silicon Valley, where such awesome advisors are easily found. Some have come through personal contacts, others through networking or LinkedIn. It is really amazing how many talented people there are in the SV community. I am glad that we get to have some of the best and brightest over every Thursday for some magic.

Email and Calendaring for a Startup

2007-08-26T16:51:04-07:00

As a startup, we don't have time or money to waste on setting up complex infrastructure. Mail servers are notoriously complex. At DataScaler, we wanted to hit the ground running, but also have an email system that would grow with us. Shared calendaring was also a must-have. Even with only four people, it's amazing how many schedule conflicts can arise without shared calendaring.

We found Google Apps to be a great fit. It uses the familiar Gmail user interface and has solid calendaring. Everything resides under the datascaler.com domain, so our email addresses are all @datascaler.com. At only $50 per user per year for 10GB of email storage, it's served us well. It's never been unavailable or had any hiccups. We can easily invite each other to meetings and manage email.

We also use the Docs & Spreadsheets app from time to time. It's not exactly Microsoft Office, though. It is a bit slow to use over the Internet and cannot import certain documents. We use it mostly for notepad-type items, but we'll likely move those over to our wiki. At a prior company, we used a hosted Microsoft Exchange system, but almost 10x more expensive. After years of using MS Exchange, I don't miss it at all.

We're an all-Mac shop, so we use Spanning Sync to keep our Google Calendars in sync with iCal on our MacBook Pros. This has worked well for us and the product is also quite inexpensive at $25 per year or $65 forever.

Another key advantage to using Google Apps is that it requires no VPN. Since our email is on the internet, it's always available from anywhere. Between Google Apps, Amazon Web Services, Skype, eFax, Unfuddle, and a few other apps, we are able to run our business quite well without any servers in our office. We never need a VPN and we don't have to purchase, house, power, and cool a bunch of servers. As a frequent business traveler, I have come to loathe VPNs, which never seem to be able to connect when you need them to. The Web truly is a wonderful thing.

Networking is Crucial

2007-08-20T13:51:38-07:00

I've been amazed at the quality and quantity of networking the DataScaler team has been doing. We decided to start networking in earnest about 2.5 weeks ago. In that time, we've met with 30+ individuals with many more to come. Almost every lunch and breakfast is booked with someone. We've targeted several types of individuals:

Sales people who can help us get product validation meetings with customers
Technical people who can act as advisors and eventually employees
Strategic advisors from the industry we are targeting
Professors and PhD students doing research in our area

When you add in the meetings with potential customers, we end up spending a lot of time talking to people. This has been amazingly helpful. While this reduces the amount of time we have for product development, it radically increases our chances of creating a product that customers will actually buy when it is released. So many startups have the "if you build it they will come" philosophy. We reject that concept and instead believe we have to pre-sell the product while it's being developed. This is working really well so far. It's great to see other people get excited about the DataScaler vision.

Skype As An Office Phone System - Part II

2007-08-19T21:35:38-07:00

As Murphy's Law would have it, just after I blog about how great Skype is as an office phone system, Skype has a big outage. This definitely affected us at DataScaler. We were unable to make or receive calls on our "office phones." The impact was lessened by two factors:

We are still pre-release, so we didn't have any customers calling us
We all have cell phones.

Despite the above, we were annoyed not to be able to use our phones. I had scheduled a vendor to call me on my desk phone, and of course he couldn't get through. Eventually, he checked his email and called me on another number.

So, what have I learned from this? I have decided:

We'll get some (not many) "real" phone lines when we move into our new office in a few months
We'll connect those lines to an Asterisk PBX and still use soft phones and our Plantronics wireless headsets. I have loved not having to type numbers into a desk phone. Soft phones are great, Skype or not.
We'll still use Skype for most calls. Only Sales, Support, and a few other key inbound numbers will be on the "real" lines. For most employees, if the phones are down, it's usually a good thing: No one can interrupt them with a phone call.

So, our experiment with Skype is still a net win; we just won't rely totally on Skype for all our office phone needs.

Skype As An Office Phone System

2007-08-10T13:53:24-07:00

We use Skype as our phone system at DataScaler. It works surprisingly well. Each person has a Plantronics CS50-USB wireless headset. These high-quality headsets play well with Skype and allow us to wander away from the desk to take calls. Using SkypeIn and SkypeOut, it's nearly indistinguishable from having a "real" phone. We save a ton on long distance fees, phone lines, a PBX, and desk phones. The headsets aren't cheap (~$180), but that's all you need. A good desk phone will run at least that much if not 2-3x more.

At a previous company, we spent $60,000 on an Avaya PBX and phones for 40 people. At a subsequent company, we spent $15,000 on an Asterisk-based system for a similar number of people. We thought that was a killer deal. Skype will cost us 1/100 of that and we don't need to pay for phone lines.

We also use the Skype Business Control Panel to allow me to centrally manage all the SkypeIn numbers and put SkypeOut credits in the individual accounts. It works very well. My only complaints are:

The Skype website is hard to navigate. It took me a long time to find what I needed to get started.
I couldn't use a credit card to pay. I wanted to buy a large block of SkypeIn numbers, so I needed to spend $200 or so. In order to do this, I had to open a Moneybookers.com account, then transfer money into it, then transfer the money to Skype. This process took several days and was a pain. Apparently, they do this to reduce fraud.

We don't yet have true PBX functionality (dial by name, call transfers, etc.), but we intend to use a managed service on top of Skype to provide that. More on that when we roll it out.

Goals and Focus

2007-08-07T15:53:10-07:00

We have some important goals to hit this quarter. In order to do that, we have a system of quarterly, weekly, and daily goals. Here's how it breaks down:

Quarterly Goals: Each DataScaler employee has a big quarterly goal that is crucial to the business' success. For the technical team, it relates to technology development. For myself and the business team, it relates to some specific customer-facing activities.

Weekly Goals: Each person breaks their quarterly goal into specific weekly goals. These weekly goals include some demonstrable end result: a technology demo, a summary presentation of customer research, etc. On Fridays, we have "Demo Fridays" where we each present our demo/presentation related to the week's goal. There is a weekly prize for the best demo, as voted on by the group.

Daily Goals and Reports: Every day before leaving, each person takes 5-10 minutes and sends a status report to the team. This report focuses on achieving 1-2 goals per day that lead up to the weekly goal. It is really easy "get a lot done" without actually moving the ball toward the goal line. We feel it is far more important to get 1-2 significant things done each day than 10 insignificant ones. Thus, we limit the daily goals to 2. By reporting daily on progress, we stay really focused and have the added benefit of increasing communication across the company. The daily report template looks like this:

This Week's Goal
<goal>

Previous Daily Goals (no more than 2)
1- <goal>
2- <goal>

Today's Achievement Toward Goals (no more than 2)
1- <achievement toward goal #1>
2- <achievement toward goal #2>

Other Achievements Today (as many or as few as you like)
1- <achievement>
2- <achievement>
3- <achievement>
4- <achievement>

Goals for Tomorrow (no more than 2)
1- <goal>
2- <goal>

This system of quarterly, weekly, and daily goals has worked really well for us so far. I'll let you know in a quarter how it's holding up.

Home Sweet Home

2007-08-06T17:25:39-07:00

After founding and funding DataScaler, we needed to find our first home. We wanted:

~1000 square feet
6 month lease
A large open area with 2 private offices for making phone calls, etc.
Running water and bathroom in the suite (this was harder to find than we thought it would be)
Mountain View, Sunnyvale, or Cupertino location

We started our search on Craigslist. We found various properties, but none that fit all 5 criteria. Since were looking for a small office and a short term lease, I figured the large commercial real estate brokers would not be much help. However, I was wrong.

We were about to compromise and go with an office suite where we'd have to go outside the building with a key to use the bathroom. Luckily, Chase Lyman and Steve Henry at Cornish & Carey found us a great office. We got a beautiful 1000sf office in Mountain View with a conference room, a kitchen, two bathrooms, lots of light, and a six-month lease. Chase and Steve really came through for us.
We'll move into a much larger office soon, so they are smart to do a good job for us on this initial office, even though they made little money on it.

To furnish the office, I gave Derek Taylor, our Director of Hardware Engineering, a budget and told him to go wild. ;-) Using a combination of Craigslist, furniture liquidators, friends, and an estate sale, Derek was able to get us desks, chairs, bookcases, tables, couches, a refrigerator, whiteboards, a projector screen, and some other goodies for about $1800. He did an outstanding job. I snapped a few photos with my Treo:

We went with 30" monitors (more on this in a future post):

One of our investors even donated a nice conference table and chairs:

We love our office; it's nice to have a great place in which to change the world ;-)

The Magic of the Valley

2007-08-04T19:59:07-07:00

Silicon Valley is unlike any other place on earth. The weather is great, the hills are beautiful, but most of all, this is the center of the startup universe. Why is that? Paul Graham and others have written about their theories on the subject. I don't pretend to know the "real" answer, but here are one man's observations. Each of these will likely be posts of their own in upcoming weeks.

Silicon Valley has:

Abundant venture capital
A risk-taking culture
Awesome startup legal expertise
Experienced startup executives
Tons of customers, large and small, within a 30 minute drive
Several major research universities that attract brilliant people
Great companies that attract brilliant people (which we then recruit away...)
An amazing network of people involved in technology, entrepreneurship, and business

These elements are stronger here than anywhere else. This is where dreams are made.

Boot 'em!

2007-08-04T16:05:06-07:00

Don Jaworski, VP of Product Development at Brocade gave us some good advice yesterday. The story starts with a band...

Don plays drums in a band, and has played in many, many bands dating back to high school. When he and his partners formed this band, however, they made a list of verboten behaviors (negativism, flakiness, unprofessional behavior, etc.) No matter how great a band member was, if he/she displayed any of these behaviors, they would "boot 'em." As a result, this band is awesome. They have a defined culture and aren't afraid to enforce it swiftly. Don encouraged us to immediately "boot" anyone who wasn't fitting in.

When choosing our founding team, I knew we could only have people who were brilliant, and more importantly, had a can-do attitude. Negativity and pessimism have no place in a startup. By definition, startups are doing the impossible. In larger companies with established customers, negativism can be healthy and pragmatism can be useful. However, when you are changing the world, comments like, "That can't be done," etc., aren't going to help. I am extremely pleased with our team at DataScaler. We're making the impossible happen.

Mission Statement

2007-08-04T11:17:05-07:00

Welcome to my personal blog. Here I will chronicle my adventures as a Silicon Valley startup CEO. My intentions with this blog are twofold:

Keep my friends, associates, and family up to date on the DataScaler adventure
Share my experiences, both good and bad, that might be useful to other entrepreneurs

I hope you enjoy it.

Chad

Disclaimer: This is my personal blog and I am personally and solely responsible for its content. Nothing posted here is an official position of DataScaler, Inc.