Building your own storage startup: Whatever you do, don't let lead dev be CEO

Product's what we want from you...


Part Two You like the idea of “doing” a startup, but what about actually starting a startup? You can see the end point - success, hopefully - but what of the hoops and hurdles?

In part one, I looked at funding, executives and staff. But now you’ve got the money and the people, what happens next? What about team composition and product?

The subject of who should comprise your core team is a hotly debated one. It's worth pointing out that I'm just a Canadian nerd and I haven't gotten rich doing any startups. Other people who have actually played the game (and won!) disagree and their views should be considered.

Jeff Ready, CEO of Scale Computing, offers the following thoughts: "On your original startup team of three, there is no sales person. Perhaps this is implied in your Gladhander, although you already have them doing quite a bit. I would take a salesperson over a tech marketer in that initial group all day long. One, because I can write the marketing stuff myself if I need to, but more importantly because customers are the key to the future funding."

John Smith, (pseudonym I'm using in this series for the marketing director of the largest tech companies in the world) offers similar advice: "The biggest one is that I am not convinced of the criticality of the tech writer in the formative stages. In my experience, the key is the Kickass Developer and the Gladhander, the latter of whom serves as the tech writer early on. In order to keep things skinny, the Gladhander typically does the schmoozing and creates the messaging and similar related content for the investors while the Kickass Developer focuses on the plumbing."

Here I feel I must poke my head in to these excellent comments. They aren't wrong: sales is absolutely critical. In fact, I view this as a key function of Gladhander: schmoozing those as need to be schmoozed and gladhanding the high rollers.

I think a CEO who can talk Venture Capitalist, sell things and crack the whip over the employees is a much easier-to-find combination than a CEO who can talk Venture Capitalist and write decently. I've encountered plenty of CEOs who could sell snow to an Inuit. On the other hand, I literally wouldn't have a roof over my head if most CEOs (or marketing VPs) were intelligible at the keyboard. Converting CxO into human is my day job.

An anonymous senior vice president at a one storage startup, who I will call John Johnson, had this to add: "There are almost always at least a couple developers. The other key to the early core of the company is you need people with cross-functional skills. Product manager / developers / content creation (slide decks are critical) / pitch person / business development (relationship person with the rolodex) / channel expert (most startups these days work with channel partners) skills all need to be included in that core team."

Trevor Pott's Desk

Only the best of desks for the CEO

Ugly babies

Do not let Kickass Developer be CEO. Do not do. By this I am not advocating that CEOs can't be engineers or developers, but rather that the CEO can't be the lead developer of the product this startup is working on.

The problem is that a huge part of the early life of a startup is seeking and incorporating feedback. Many experts I reached out to pointed out that they would seek out potential early customers before ever considering putting together money for a startup and this must continue for years if success is to be achieved.

This process of engaging with customers will lead to changes in the design of the product. If the product is the CEO's baby, there's a problem. Nobody likes being told their baby is ugly and when the person in charge is emotionally invested in a specific solution, startups usually fail.

Again, there are exceptions. Dheeraj Pandey, CEO of Nutanix explains, "I couldn’t relate to the Gladhander. As a CEO, I wrote a ton of code. Every partnership is different, you know. Just like a family – parents, siblings, etc, and making money is the last reason why people build sustainable companies. The biggest reason relates to the “right brain” – autonomy and an elusive desire to change the world."

As one storage VP puts it: "the delicate balancing act of early startup life. Too much passion and there is an argument and a critical team member leaves. No enough passion and the plane flies into the side of the mountain. Too much debate and the product never ships. Not enough debate and the plane flies into the side of the mountain."

With the who and how of starting our startup sorted, let's take a break from people and politics and dive into what the potential technology of our startup might look like.

Tech 1.0

Before any of the funding or shenanigans above, you need an idea for a product. As we're designing a storage startup, I'll lay out a storage product. To design this product, I'm mooshing together ideas from dozens of storage startups alongside my own, so I'm sure lots of folks will see something of their own startup reflected in this design.

In a perfect world we want to build a storage solution that meets all needs for all people. Everyone else will simply stop using the competition because ours is so awesome. Boiling the ocean, however, never works as a starting proposition, so we'll focus on one primary problem.

NVMe is all the rage, and faster is typically better, so the core goal will be to build a storage product that goes really fast and takes advantage of what NVMe has to offer. NVMe is good for fast storage primarily because of high queue depth.

Let's start with the new Intel Broadwell-based Xeon-Ds with integrated FPGA. Customize the FPGA for your high I/O workloads. This can help get around the part where NVMe is really, really CPU intensive. Get a NIC that can handle offloading without imploding in order to also drive CPU usage down. Pack it all into one of Supermicro's 48 drive NVMe chassis and slap your brand on it.

As part of your sexy software you build a scale-out storage solution similar to Coho or Solidfire so that you can handle not only failure of the device, but failure of (at least) a node. Let's say it's object based and allows for n copies of an object.

Qualify a good switch as part of the solution, one that can handle multicast really well. Use multicast for transmitting reconvergence data and other inter-node communications. This makes rebuilding of nodes easy and low-impact, even with huge NVMe setups. The rebuild stress ends up shared equally across the cluster, but the whole of the cluster doesn't have to play silly buggers trying to figure out who is doing what. The good switch is necessary because cheap switches are pants at multicast, and you're going to need every scrap of networking you can eke out.

Build locality awareness into the storage platform and make sure any node can serve up storage. If you have a 32-node cluster, use 16 virtual IPs so that when you assign storage to a requestor they can not only manage to load balance requests across the whole of the cluster, but the cluster can make sure that the assigned storage has replicas physically on the nodes responding to the storage request.

Generic Storage Startup Timeline

Most of this can be done by welding together various open source projects. As a 1.0 to prove viability to an angel investor that very well might be good enough, assuming you can demonstrate that you have identified pain points and have a solid roadmap for replacing some (or all) components by A round. You need something to patent, and the investor doesn't want to pay you to make a startup anyone can come along and clone.

Be wary of other startups! As I discussed above, this design – like every storage startup – is taking elements from various existing companies' products*. Make sure you know who has patented what, and you work around that. Get Gladhander to earn their keep!

Knowing what you want to build, however, is not enough. You need to be able to build it in stages, monetize it at different points, and have some idea of how you will handle investors, staff, partners, community and politics.


Other stories you might like

  • It's 2022 and there are still malware-laden PDFs in emails exploiting bugs from 2017
    Crafty file names, encrypted malicious code, Office flaws – ah, it's like the Before Times

    HP's cybersecurity folks have uncovered an email campaign that ticks all the boxes: messages with a PDF attached that embeds a Word document that upon opening infects the victim's Windows PC with malware by exploiting a four-year-old code-execution vulnerability in Microsoft Office.

    Booby-trapping a PDF with a malicious Word document goes against the norm of the past 10 years, according to the HP Wolf Security researchers. For a decade, miscreants have preferred Office file formats, such as Word and Excel, to deliver malicious code rather than PDFs, as users are more used to getting and opening .docx and .xlsx files. About 45 percent of malware stopped by HP's threat intelligence team in the first quarter of the year leveraged Office formats.

    "The reasons are clear: users are familiar with these file types, the applications used to open them are ubiquitous, and they are suited to social engineering lures," Patrick Schläpfer, malware analyst at HP, explained in a write-up, adding that in this latest campaign, "the malware arrived in a PDF document – a format attackers less commonly use to infect PCs."

    Continue reading
  • New audio server Pipewire coming to next version of Ubuntu
    What does that mean? Better latency and a replacement for PulseAudio

    The next release of Ubuntu, version 22.10 and codenamed Kinetic Kudu, will switch audio servers to the relatively new PipeWire.

    Don't panic. As J M Barrie said: "All of this has happened before, and it will all happen again." Fedora switched to PipeWire in version 34, over a year ago now. Users who aren't pro-level creators or editors of sound and music on Ubuntu may not notice the planned change.

    Currently, most editions of Ubuntu use the PulseAudio server, which it adopted in version 8.04 Hardy Heron, the company's second LTS release. (The Ubuntu Studio edition uses JACK instead.) Fedora 8 also switched to PulseAudio. Before PulseAudio became the standard, many distros used ESD, the Enlightened Sound Daemon, which came out of the Enlightenment project, best known for its desktop.

    Continue reading
  • VMware claims 'bare-metal' performance on virtualized GPUs
    Is... is that why Broadcom wants to buy it?

    The future of high-performance computing will be virtualized, VMware's Uday Kurkure has told The Register.

    Kurkure, the lead engineer for VMware's performance engineering team, has spent the past five years working on ways to virtualize machine-learning workloads running on accelerators. Earlier this month his team reported "near or better than bare-metal performance" for Bidirectional Encoder Representations from Transformers (BERT) and Mask R-CNN — two popular machine-learning workloads — running on virtualized GPUs (vGPU) connected using Nvidia's NVLink interconnect.

    NVLink enables compute and memory resources to be shared across up to four GPUs over a high-bandwidth mesh fabric operating at 6.25GB/s per lane compared to PCIe 4.0's 2.5GB/s. The interconnect enabled Kurkure's team to pool 160GB of GPU memory from the Dell PowerEdge system's four 40GB Nvidia A100 SXM GPUs.

    Continue reading
  • Nvidia promises annual updates across CPU, GPU, and DPU lines
    Arm one year, x86 the next, and always faster than a certain chip shop that still can't ship even one standalone GPU

    Computex Nvidia's push deeper into enterprise computing will see its practice of introducing a new GPU architecture every two years brought to its CPUs and data processing units (DPUs, aka SmartNICs).

    Speaking on the company's pre-recorded keynote released to coincide with the Computex exhibition in Taiwan this week, senior vice president for hardware engineering Brian Kelleher spoke of the company's "reputation for unmatched execution on silicon." That's language that needs to be considered in the context of Intel, an Nvidia rival, again delaying a planned entry to the discrete GPU market.

    "We will extend our execution excellence and give each of our chip architectures a two-year rhythm," Kelleher added.

    Continue reading
  • Amazon puts 'creepy' AI cameras in UK delivery vans
    Big Bezos is watching you

    Amazon is reportedly installing AI-powered cameras in delivery vans to keep tabs on its drivers in the UK.

    The technology was first deployed, with numerous errors that reportedly denied drivers' bonuses after malfunctions, in the US. Last year, the internet giant produced a corporate video detailing how the cameras monitor drivers' driving behavior for safety reasons. The same system is now apparently being rolled out to vehicles in the UK. 

    Multiple camera lenses are placed under the front mirror. One is directed at the person behind the wheel, one is facing the road, and two are located on either side to provide a wider view. The cameras are monitored by software built by Netradyne, a computer-vision startup focused on driver safety. This code uses machine-learning algorithms to figure out what's going on in and around the vehicle.

    Continue reading

Biting the hand that feeds IT © 1998–2022