Kacin Point

Cisco Validates Appliance-based Software Delivery

A lot of people were interested in the January 4 announcement that Cisco is purchasing IronPort for $830 million (well, at least they were interested in IronPort news before the iPhone argument between Apple and Cisco got in full swing).

IronPort, a five-year-old, privately held company, makes appliances for Web security and management. From our perspective, this deal certainly seems like a big validation that appliance-based software has an important place in enterprise software ecosystem.

But this notion of appliance-based delivery still seems to be flying under the radar a bit.

In an unscientific search, I found:
- 500 stories on Google News on the Cisco, IronPort acquisition.
- 114 of those stories mentioned anything about an appliance.

If we’ve learned anything this week, it’s that KACE is onto something with Appliance-based Software Delivery (AbSD). It’s just a matter of time before the press realizes how big this appliance thing is.

Mixed Environments: Making multiplatform networks easier to manage in the mid market

For companies whose IT networks include a lot of diverse devices running on different OS platforms, managing a mixed environment poses a good amount of trouble for the IT professionals involved. EMA analyst Andi Mann wrote an illuminating article for Network World a couple of weeks ago where he talks about this problem—how to combine the flexibility wanted by business users with the standards needed by IT managers, and make it all accessible and centrally manageable.

Some larger enterprises can absorb the huge investments of time and money required for a tool-juggling act and/or an overly complex enterprise solution. But the mid market has been largely underserved by IT automation tools, and been forced to make do with downscaled (read: virtually stripped of essential features to cut their expense) solutions to fit their smaller budgets and node requirements.

So it boils down to making a choice between user flexibility and standardization to ease network management—a choice that’s always going to leave someone (either users or IT managers) underserved and overworked.

At KACE, our focus has always been the mid market, and we’ve just launched the newest version of our comprehensive IT network management solution for SMBs—KBOX 1000 Series Systems Management Appliances, which supports just about any network addressable device running Windows, Mac OS X and Linux. And because it’s a plug-and-play appliance with an especially user-friendly Web-based interface, there’s far less complexity around deploying and using it.

You can get a live demo that shows how easy it is to deploy and use KBOX 1000 Series appliances here.

Bringing automated IT provisioning to the mid-market was no easy task

At KACE, we hear from our customers time and again that provisioning is a big headache for them, and I can relate. The reason we started KACE in the first place was to create IT automation solutions for mid-market companies, many of which rely on one central IT location to service a dispersed network of servers, desktops and laptops. Manual provisioning in these situations can be a real challenge.

We’ve just gone to FCS (first customer ship) of our new KBOX Series 2000 Systems Deployment Appliance to address the automated, remote provisioning challenge. So it would seem that now is a good time to reflect on the engineering effort this took. Interestingly, we found the work was less about writing code and more about digging into and understanding the esoteric, low-level problems of systems deployment and figuring out ways to solve them. It turns out that much of what the appliance does (remote, agentless, unattended scripted installation, emergency repair, and more) is just plain hard to do and that's what made the development a real challenge.

We built the KBOX 2000 from the ground up to support network boot and deployment management services over the wire, requiring a detailed integration of PXE, DHCP, device drivers, and custom OS boot environments. PXE is kind of cryptic and matching up device and network drivers is always a pain, as is making sure the DHCP and boot ROM locations resolve. All of these components are known to be testy, and we ran into more than a couple of problems getting them sync’d up.

We had a lot of frustration for a while there: PXE boot was fine, ram disk tftp transfer was fine, but the bootloader failed with "Error 21” and a hex message. It worked fine on some machines and but others gave that failure message. It turns out NTLoader.exe for Win XP only would boot completely on certain versions of PXE boot ROMs on specific hardware. We used network traffic analyzers to verify boot packets were intact, but we couldn’t figure out why some PXE boot ROMs didn’t work with the NTloader version. After days of pulling out our hair, we swapped in the 2003 NTLoader for the Win XP machines and then, finally, it all just worked.

There were a bunch of things that we did in engineering that will really help the KBOX 2000 users. We designed the KBOX’s disk imaging capabilities to cleverly boot, over the network, a target-managed machine in a RAM based OS. Whether the OS is Windows or Linux or other, the machine is booted into ram, the KBOX mounts the local disks and performs the imaging operations. While booted into the memory, the KBOX streams the disk images back and forth between itself and the managed machine.

The KBOX actually uses that RAM disk to clone windows formatted drives. It allows not just full sector-by-sector disk capture, but it also recognizes NTFS specific partitions. When NTFS is identified, the KBOX can optimize the image capture so it doesn’t have to copy all sector information but just the part of the disk being used. This, along with image stream compression allows for both smaller image data footprint in the storage archive and far less data traveling over the network.

We enjoyed engineering the new KBOX not just for the mental challenge of it all, but because we knew mid-market IT admins are looking for a systems deployment appliance that could solve the pain of manual deployments without adding a lot of expense and complexity to the mix. I’m proud to say I think we’ve done that with the KBOX 2000!

Enterprise software delivery: Is the hype of SaaS worth the risk?

Software as a Service (SaaS) is being hyped in the media, which is odd because SaaS can be so unreliable. Just ask the frenzied sales teams who tried to access critical SalesForce.com data last December—just days before the end of the quarter. These sales pros were rebuffed most of the day due to outages.

Sure, SaaS looks pretty good on the surface to an IT buyer. Cheap. Easy to use…

I was talking to Maryfran Johnson, editor of CIO Decisions, recently. Maryfran spent years at Computerworld, and for the past year or so has been managing CIO Decisions, which she said targets the "Fortune 50,000"—all those mid-market companies that are often overlooked by mainstream IT media.

We were talking about the pros and cons of SaaS—she had heard from so many CIOs of mid-market companies that they want simplicity, and that's what SaaS is supposed to deliver. My point back to Maryfran was that appliances take simplicity to the next level.

IT and business people want automation solutions that maximize control, minimize time/effort, and are cost effective. Appliances done right maximize the output of this equation in both the short term and long term. I think SaaS makes a short-term superficial marketing attempt to solve this equation.

The problem is that SaaS is a one-size-MUST-fit-all application, so it can’t be tuned to individual end-user specs or should I say individual “end-Company” spec. SaaS doesn’t offer the end user (the corporate customer) any flexibility or choice in service levels. Vendor-dictated downtime for upgrades and maintenance means that entire end-user communities will experience service outages over which they have no say or control. Slow WAN connections and usage spikes can affect performance. And the security risks that arise when a company allows external control of internal data can add to the overall costs caused by all of the above.

So SaaS requires that the customer take a leap of faith that isn’t rewarded in low cost, scalability, reliability, performance or security.

So why all the hype? I’m not a marketing guy. I have no idea. SalesForce.com markets the idea that “no-software is best”--if that’s the case, what is running their huge operations centers and CRM applications? ☺

When we founded KACE, we spent a LOT of time evaluating our different software delivery platforms, including SaaS. After some serious research and thought, the answer to us was clear: we went with an appliance-based delivery model (we call it AbSD). 

AbSD trumps SaaS by offering a pre-configured, pre-tuned, made-to-spec solution, which means built-in simplicity for the end user. The appliance is an integrated hardware-operating system- server app that’s plugged into the network for immediate use. AbSD gives the end user support from a single vendor for all of its components, and the appliance can automatically retrieve vendor updates for the application(s) and integrated operating environment. Centralized management enhances reliability and security since only the OS components necessary for the application are exposed to users.

For startup software vendors considering their options for a delivery method for their new products, we here at KACE say ignore the hype and go with the delivery method that works the best at the lowest cost: go with an appliance.