Shaping Technology Strategy

CTO Journal

Subscribe to CTO Journal: eMailAlertsEmail Alerts newslettersWeekly Newsletters
Get CTO Journal: homepageHomepage mobileMobile rssRSS facebookFacebook twitterTwitter linkedinLinkedIn


CTO Authors: David H Deans, AppDynamics Blog, Louis Nauges, William Schmarzo, Shelly Palmer

Related Topics: CTO Journal

CTO: Article

Brand New Linux Distro Expected To Be a "Solaris Killer"

CTO John Buswell Describes It As "Unlike Anything Currently on the Market"

A stealth start-up in Athens, Ohio whose name is Spliced Networks LLC is on the threshold of announcing a new Linux distribution that its young CTO John Buswell describes as "unlike anything currently on the market."

It will supposedly eliminate "bloated package management," allowing for upgrades or rollbacks in less than 30 seconds. To compete, Red Hat, Novell and Mandrakesoft would reportedly have to "completely re-engineer their solutions away from RPM and other package management systems."

The company thinks - allowing for market reaction - that it "could spell the end of server operating systems as we know them today, and would likely put an end to Solaris."

The Spliced distribution is supposed to upgrade in 15 second without rebooting and roll back in seconds too. Applications supposedly upgrade in seconds with minimum interruption to services.

Buswell says they're working on the next-generation of appliances - both software and hardware.

Oh, heck, let them tell their own story. This is what his e-mail says:

"The appliance market was the big rage just before the dot.com bust, they offered fixed-function devices, typically 1U or 2U, that you could set up in 15 minutes. Things like Sun's Cobalt RAQ, and devices from companies such as CacheFlow. At one point in 2001, I think every major hardware vendor had jumped on the appliance bandwagon, but by 2002, a lot of they were starting to drop the idea, in favor of more general-purpose servers.

"Spliced Networks was founded in 2002, right around the time the appliance market was starting to dwindle. Our goal was to make Linux installations easier to manage, easier to deploy and provide a built-in framework that would help businesses deploy scalable networks, even if they didn't know what they were doing. We spent about a year doing R&D on small things, not major earth-shattering innovations, just small things that make improvements.

"Appliances have some excellent things going for them, they typically have a small secure embedded OS, that OS typically runs from flash memory (no disk failures), they are easy to manage (you need no knowledge of the underlying OS, you just need to have some idea of what you're trying to deploy) and they produce nice fancy stats. Appliances have downsides too, they are typically fixed hardware, your network grows, you have to replace the appliance (some of these appliances come with a $20k price tag), if a new vendor comes into the market with a better device, and your vendor doesn't come up with comparable features you want, then that's a $20k box you have to offload, and a new $20k box you have to buy.

"A quick scan of eBay, and you'll see lots of liquidators selling $20k+ appliances for under $500, a testament to the lack of scalability and why companies reverted back towards general-purposes servers.

"Now, general-purposes servers are more costly to administer, you typically need a good administrator, knowledgable in the OS, security and the application side. General-purpose OSes like Red Hat Enterprise, SuSe, Linux Mandrake, Solaris and Windows 2003 Server are just that general-purpose. With Linux, you can spend hours doing rpm-e just to remove stuff you don't need on your web server. This is a lot of unnecessary work. If you look at a lot of dedicated server providers, you'll find hundreds of Red Hat, BSD and other servers that haven't been updated in years, that are compromised, even those running telnet on open ports. Why? Well typically some small business has purchased a dedicated server, paid a web developer to upload their site, and have just left it there. The dedicated server provider has just provided the box and expects the user to admin it themselves. Sooner or later that box is compromised and sending spam and worms to the general public.

"This is where we come in. The reality is that most servers only handle a couple of 'applications.' Larger companies and ISPs typically dedicate server(s) to a specific task such as http, https, ftp, smtp, ids, etc. Even very small companies only share two or three applications on a single server. So why install the kitchen sink (e.g., RHEL), when you really only need smtpd, pop3d and dns running on your server?

"What we have done is create a couple of new things. First we have a project called LIMBS (Linux Image Management/Boot System) that places a layer between the kernel and the init process. This layer boots the system much like a router or switch from a single flash image file. This image file is placed on a solid-state memory device that replaces the hard disk in your server. The use of solid-state memory eliminates the HDD as a point of failure. HDDs are moving parts devices and will eventually fail. Solid-state memory, at least the stuff we're using, is extremely fast, and boots the system very quickly.

"Once LIMBS loads the image, and manipulates the boot process, it hands off control to the init process. Here we have completely replaced the traditional init system with our own. Our init system uses a common configuration file to generate the various configuration files used in Linux. A Linux distribution can have a lot of configuration files, all in different formats, the idea here is that a common configuration format lets the user configure aspects of the OS without knowing the OS at all. What's better is that we have a web-based interface (with CLI and SNMP support to follow) that lets you manage the system, which will generate the config file for you so you don't even have to know the format.

"Finally, our init system will load one or more Application Specific Linux (ASL) images. These are similar to router/switch images, but instead of providing the OS, they provide the application. These ASL images are important, because they simply extend the common configuration format and web interface so when you load an ASL image the ability to manage that image is available through a common interface. These ASL images are also key to providing enhanced security, it'll be more apparent when we make the release announcement, but the ASL images are highly secure. If someone was able to compromise the application, they cannot manipulate the config or data the application has access to nor can they manipulate the OS. This is a capability that has been designed into the system.

"We will be starting an initiative program that allows OEM and VARs to build 'Open Appliances' that conform to specific quality and performance guidelines. The idea is that you should be able to buy a certified Open Appliance from Company A with our Linux OS, buy a DNS ASL image from Company B, and a Firewall ASL image from Company C, and have them all work seamlessly on the same box, with the same management interface, as if the solution came from a single provider. It's essentially taking the best aspects of the old appliance devices and combining them with the openness of a generic server.

"To give you an idea of the potential, it essentially places the power back in the hands of the user. Today, lets say Company A has a great SSL Acceleration product, and Company B has a great SSL VPN product. Company A and Company B are major competitors, you'll never see them partner, so your only choice is to spend $$$$ buying both boxes. What we want to see with our release, is Company A making their software available, and Company B making their software available, and you buying hardware from Company C that partners with both A and B. The end result is that you have the best SSL Acceleration and the best SSL VPN integrated into the same box. You're shipped a box with 2GB RAM, your network load increases, so you can go anywhere and buy another 2GB RAM to max out the memory in the system, you're not stuck overpaying $$$ to some vendor for a new box or the same memory with a crazy price tag.

"Anyone can write an ASL image, we have developed a key based system that allows vendors to provide demos, time-limited licensing, and feature-licensing capabilities to their ASL software. There really isn't much difference between producing an ASL image and an RPM. There is a little more work involved with ASL, but the security and management advantages are worth the extra mile.

"What we're really hoping for with this initial release is to help increase the pace at which Linux is deployed. With our solutions, an OEM/VAR can deploy Linux without any Linux-knowledgeable staff on hand. You don't need Linux knowledgeable staff for our solutions, thinking that you do would be like thinking you need VxWorks-knowledgeable staff to deploy Cisco solutions. I think that we'll see a lot of companies that are traditional Microsoft shops, who are losing out to Red Hat and other deployments, think about jumping ship, and offering our solutions to retain customers. That's how we hope to get our foot in the door at many new vendors who would traditionally not consider Linux.

"It should also hopefully put an end to particular vendors who charge $10k+ for appliances that are no more than cut-down versions of Red Hat running Bind with a bad web interface. One of the things that really annoys me about the IT industry is that often, good and innovative companies fail because better-funded companies with strong sales/marketing teams can outspin them. These are the types of companies who want an 'unknowledgeable' customer they can wine and dine and convince that $10k is worth it for a $1,000 Linux box and an interface any college student could whip together in PHP in a couple of weekends.

"As a small company, we're currently building strong VAR and OEM channels across the Americas, APAC and EMEA. I'm originally from Ireland, so I'm lucky enough to have a lot of industry contracts in Europe, as well as here in the US and APAC, that has allowed us to develop channels quickly in these regions. We hope to start adding a lot of new ISVs, as soon as we release and those companies see the advantages of using our OS for rapid time to market. Our ISV program, which we're launching next week, should help even the smallest companies compete worldwide for business. So hopefully, we'll see a lot of new Linux-based start-ups using our OS, and creating ASLs for new web-based applications (such as Google maps). Our solutions make it very easy for even a single developer with a couple of weekends and a few hundred $$$ to start a business in a niche market. For example, by joining our ISV program, a small company who developed a web-based chiropractors' patient management system can market that ASL worldwide, our OEM/VAR partners can integrate their ASL with our OS on the VAR/OEM's hardware and sell a finished box to the end users.

"From a business perspective, we have a unique business model. Everyone here telecommutes, we use broadband and OSS technologies to conduct business and development on a day-to-day basis. We have a 'no suits' policy, our engineering team shares in the non-technical business tasks such as pre-sales, support and even marketing. It has had interesting results. More importantly our customers are going to get a straight answer rather than a marketing spin. I think in the end, businesses will find that refreshing, especially in a market when published stats are often 'theoretical' and sales teams make 'accidental' claims of functionality their products actually cannot do. It also eliminates the games of phone tag you need to play in order to talk to the person to get your problem resolved.

"We will also be able to offer our customers indemnification in regards to our products and any intellectual property issues that might arise. Probably one of the more interesting business-related things we are doing is for our premium OEM/VAR and ISV customers. We're supporting our own products through a minimum 72-month lifecycle. Once a product is EOL'd (end of life'd), we will continue to support the current software release, fixing any serious problems and security issues for up to 60 months past the EOL date. We also guarantee that we will not EOL a product for at least 12 months from the date of first release. For our premium OEM/VAR and ISV customers, we will support their products under the same lifecycle strategy in the event the OEM/VAR or ISV goes out of business, abandons the software or no longer wishes to support ASLs. If their products aren't supportable, we will provide their customers with free upgrades to a comparable software product. For small vendors, its icing on the cake, for customers it's a win-win situation. Our premium partners, also receive pre-sales and post-sales support, so if their support staff can't resolve something, they have our team to fall back on.

"We're starting out with a release of the OS, some services to support OEM, VARs and ISVs, our line of AMD64-based appliances, and a couple of application solutions. Initially, this will be DNS, SMTP, and we currently have VoIP, IDS and some other stuff in beta. However, we plan to grow our portfolio rapidly over the next 12 months. We have developed a special build system that allows a very small development team to manage a large number of projects."

More Stories By Maureen O'Gara

Maureen O'Gara the most read technology reporter for the past 20 years, is the Cloud Computing and Virtualization News Desk editor of SYS-CON Media. She is the publisher of famous "Billygrams" and the editor-in-chief of "Client/Server News" for more than a decade. One of the most respected technology reporters in the business, Maureen can be reached by email at maureen(at)sys-con.com or paperboy(at)g2news.com, and by phone at 516 759-7025. Twitter: @MaureenOGara

Comments (10) View Comments

Share your thoughts on this story.

Add your comment
You must be signed in to add a comment. Sign-in | Register

In accordance with our Comment Policy, we encourage comments that are on topic, relevant and to-the-point. We will remove comments that include profanity, personal attacks, racial slurs, threats of violence, or other inappropriate material that violates our Terms and Conditions, and will block users who make repeated violations. We ask all readers to expect diversity of opinion and to treat one another with dignity and respect.


Most Recent Comments
dfarning 05/09/05 09:37:49 PM EDT

Are you a journalist or do you just forward emails and press releases people send to you?

dfarning

Peter Tribble 04/12/05 06:33:23 PM EDT

The use of Solaris in the title is clearly gratuitous and irrelevant. The article itself barely mentions Solaris, while the press release on Spliced Networks own web site takes aim at "traditional Enterprise Linux distributions".

John McLaughlin 04/12/05 09:09:02 AM EDT

A "Solaris Killer"? Does it really make sense to just repeat such unfounded claims by a new, unproven vendor?

Sun offers worldwide technical support for Solaris, including 7x24 support. Does Spliced? For many organizations which deploy Solaris, such support is required.

50-70% of Solaris systems are sold through Resellers. Does Spliced have a business proposition for Resellers that will make them more money selling a Spliced solution rather than a Sun Solaris solution? Remember that Sun has competitive AMD based workstations and servers as well as SPARC systems.

There are 10,000+ commercial Solaris apps that give people compelling reasons for buying a Solaris-based system. Does Spliced have a program to help those vendors port from Solaris to their Linux?

By the way, most of the popular free and open source software also runs on Solaris, so Spliced has to real edge there.

Many commerical and classified sites choose Solaris for its security features. Has Spliced makes the changes to Linux and gone through the process to get their configurations certified by the various government agents around the world?

Some applications scale well horizontally, some veritcally. Can you get linear performance improvements from Spliced Linux by adding more CPUs? How many? 10, 32, 64, 128, 256? That's what Solaris delivers today after hugh amounts of engineering effort on with milllions of dollars of gear, with hundreds of apps in mission critical environments over 10+ years.

mr stealth 04/11/05 07:14:40 PM EDT

"A stealth start-up in Athens, Ohio whose name is Spliced Networks LLC is on the...."

I guess they aint so stealth no more ... Once you write about them, they become un-stealth. Sounds like a neat idea. Server admin so easy even an MCSE could do it.

I guess, the question is: Has anyone really analyzed the amount of work sysadmins do on a farm with many linux systems? And if a site has lots of systems, wouldn't they also already have some form of automated management.

I imagine disk space is so cheap these days that people don't worry too much about what software is installed.

Personnally I'd prefer something where the kitchensink is available, but not necessarily activated. And for servers exposed to the internet, you would probably want to keep a keen eye on 'em anyways.

NICK VLAHOS 04/11/05 02:30:33 AM EDT

"GASP"...if ALL of this is TRUE!! "YAWN"...if it ISN'T!!

Behrang Saeedzadeh 04/10/05 05:28:04 PM EDT

I've not read the entire entry so forgive me if I'm asking obvious things: Is this distro going to be both suitable for servers and desktops? If so, I guess the most important feature that a desktop distro should have is not a moderner package management tool, apt suffices for me, but my newbie friends all complain that Linux is very weak in multimedia area and they leave it only for that reason. Is this new distro going to include MP3 players, for example?

a former classmate of the cto's 04/10/05 12:12:45 AM EDT

So is there anything more than vapourware available on this? Spliced networks don't even have a website. Any real technical details? Is LIMBS going to be opensourced?

I knew John Buswell in college (University of Limerick, IT degree, graduating class of 1999, although it took him a year longer). I very much doubt he has the technical ability to pull this off. Perhaps some of the computer society server admins from Limerick during that period would like to comment on his abrupt departure from the sysadmin team?

Seeing as he apparently works for Nortel at the moment (cached google link http://66.102.7.104/search?q=cache:NESQy3PGk6gJ:www.skynet.ie/~pagan/+jo...)
I'd love to know what their view is on his "start up".

Leon Brooks 04/09/05 10:45:37 PM EDT

Slack! This is just a press release in a wrapper. What is it with this younger generation and failing to do any research?

The release looks like mostly hype, but there are some interesting snippets in there.

I've had to reboot some servers twice since installing Mandrake 10.1 on them, all in the name of kernel updates, a waste of good uptime IMESHO, so escaping these reboots would be nice. Two-Kernel Monty, anyone? Fast updates and rollbacks etc can already be done in seconds with either or both of RPM and a versioning filesystem. Anybody can also set up a chrooted copy of their environment, experimentally update within that and see what happens, disposing of the evidence with similar speed should the results be embarrassing. None of the above si within the reach of MS Windows at all, BTW.

Magnus Grander 04/09/05 08:21:59 PM EDT

Welcome to the Linux world. It is always good to see new inovations. The old companys like Red Hat, Microsoft and so on. They are so big so thaey start to lack ideas and value for money. So what will you charge for your OS ?

Jason 04/09/05 06:10:43 AM EDT

The business concept is not hard to understand but how much licensing cost would be invovled. For a third world start-up buiness, this would really be something. I encourage the initiative and look forward to further news in this regard.