Once again, my good friend Rikki Endsley will be moderating the (probably) 90 minute session. They are giving preference to women in the Washington DC area (or people who are going to be attending LISA anyway), so if you're in that region and this sounds like something that interests you, email lisa13gurus@usenix.org (or drop a line here, and I'll get the message to them).

Thanks!

LOPSA-East

But I've been doing good, fun things. For instance, on May 3rd and 4th, I went to LOPSA-East, which was yet another really great conference. There was somewhere around 150 attendees this year, and it was really nice to see everyone again from previous years.

Way back in October of 2011 (were some of you even born then?), I asked about a class on SSDs, to see if there was any interest. Well, in October of 2011, the earliest I could have done it was spring of 2012, and didn't get around to finishing the course before then, so spring of 2013 it was, and I taught the SSD class on Saturday afternoon. Only three years in the making. That's cool, right? :-D

If you were in my class, you probably have the slides from the USB key. If you weren't in my class, then you'll be happy to know that since I don't really intend to teach the class again (although if my feedback is overwhelmingly positive, I'll consider it), I opted to have it recorded, and whenever that goes live, I'll be linking to it from here and including my full slide deck, too.

Storage Field Day

At the end of April, I went to Denver to do Storage Field Day. I haven't had a chance to write about the things I saw yet, but I'm very excited to talk about what we saw with Pernix Data. If you want to see some cool ideas, watch the videos there. I'll write more as soon as I get time.

LOPSA stuff

We're still in the swing of the election season. You might have seen when I updated my earlier post that the LOPSA Live transcript had been posted. That was the first of two candidate sessions. The other is tonight at 9pm, so follow the instructions by Aaron Sachs for connecting to #LOPSA-Live on Freenode and come ask the candidates good, hard questions.

The election is coming up next month. I've posted my series of discussions on internal concerns (including membership numbers, member communications, and operational transparency. Starting tomorrow, I'm going to start posting discussions related to external concerns - we have a lot of problems with marketing and how we're seen externally...when we're seen at all. Make sure to watch for those blog entries, too.

LISA Training

I haven't posted anything about it here, but I'm working with Dan Klein to help get training ideas for LISA'13. For the past several years, I've been involved as a blogger at the LISA conference (along with Ben Cotton, Marius Ducea, Greg Riedesel, and many others. I'm planning on continuing that for as long as they'll have me, but it's also nice to be able to contribute to the program in some small way, too. This means that if there's training that you think LISA should have, but doesn't, let me know and I'll do my best to figure out how we can have it.

Actual, "I get paid to do this" work stuff

At work, we've been doing all kinds of things. I've now got a production vSphere cluster, a new Nimble storage box, I'm trying desperately to get new gear for my core switch (I'm going with a pair of Nexus 5548s and six FEX to go along), and I need to order more five or six server racks to replace some of the ones we have now.

I continue to be mystified by the way that academia works. Specifically, budgeting and deadlines. For reasons that I'm unable to fathom, in order to get things on this year's budget, I have to order hardware and have it delivered and in my space by the end of June. Not, "ordered and paid for". Ordered, delivered, and in my space. I've thought about it, and I can't come up with any kind of compelling reason for this rule. Anyone with more experience in academia than I have want to weigh in? I'm at a loss.

Personal Stuff

I've finally bit the bullet and decided to get LASIK.

I'm in a large-ish metro area now, and the technology has been continually developing for a couple of decades, and I think it's matured to the point where I'm cool with people cutting my eye open and burning part of it away using lasers. I can't be 100% about technology enhancing our lives unless I walk the walk and take advantage of it, so I'm doing it.

I went in last week for my "free consultation", which determined that I was an excellent fit for normal "LASIK" surgery. If my cornea had been too thin, I guess I could have gotten either LASEK or PRK, both of which work well but have a longer healing and recovery time. Turns out my cornea is just fine.

Also, can I just say - they have the coolest eye equipment I've ever seen there. I've worn glasses or contacts since elementary school, and I've lived in a dozen cities or so since then, so I've seen my share of optometry equipment, but man, the toys the LASIK guys have are nuts. I'm practically blind, so when they said, "take off your glasses and look in this machine, and you'll see a hot-air balloon", I thought, "please, I'll be lucky to see a blurry light". Sure enough, looking into the machine, it was blurry...for a second. Then, like a camera, it "autofocused" and just like that, they had nearly my exact prescription. Awesome!

So the whole "lasering my eyeballs" thing is happening tomorrow afternoon. I honestly can't wait. I've been thinking about it for years, and having it this close is really exciting. I'll make sure to update early next week with the results.

So there you go. That's what I've been up to. I'll try to get back to posting more regularly, and maybe even on topics that you care about! Wouldn't that be exciting? ;-)

We'll see. Thanks!

You might remember that I'm running for election, as well. My candidate statement is at that link, and I've written a series of articles on my LOPSA blog expounding on the internal concerns that I feel need more concentration by the Board.

Tonight would be a great time to get a feel for how the potential Board members feel about topics relevant to our profession.

I'll be reminding people to tune in via twitter prior to the meeting, so watch for it there. I'd really appreciate your support and your questions, so please come and take part. This is your profession, so start taking part in it now.

Edit: The session has finished and you can read the transcript here:
https://lopsa.org/content/lopsa-live-may-9-2013-candidate-forum-transcript.

I have worked very hard for LOPSA as a volunteer, but there are certain things that you can only do from inside "the machinery", as it were, so I'm throwing my hat in the ring, and I'm here to ask for your support.

Here is a brief overview of my history with LOPSA:

• Joined in February, 2009
• Joined the LOPSA-NJ chapter when I moved to New Jersey
• Volunteered as the marketing coordinator for PICC10 Conference in 2010
• Founded the LOPSA-NYC Chapter in 2010
• Served as the program chair for the PICC'11 Conference in 2011
• Founded the LOPSA Columbus chapter in 2011
• Was honored with the 2011 Chuck Yerkes Award for outstanding individual contributions to online forums
• Joined the Boston LOPSA Chapter in 2012

I have prepared a detailed candidate statement. It's eight pages, explaining what I want to do with LOPSA, and where I see problems with what LOPSA has been doing, and how we need to fix it. The biggest complaint against LOPSA is that there's no value for membership, and if you read that document, you'll see that I'm going to work with the other Directors to fix that. We can't just sit on the sidelines. The future of our profession is too important, and we, as an organization, need to start directing our own future.

Because eight pages is extremely long, I am going to be breaking down my candidate statement into chunks that are more easily digested and posting them on my blog on LOPSA.org. My first entry is up, announcing my candidacy. I'll be adding more in the coming days.

Thank you for the support you've already shown me, my blog, and the various efforts that I've made on behalf of LOPSA and system administration. I don't feel like I've done enough yet, but with your support, I can do more. If you are a LOPSA member, I would greatly appreciate your support in the coming election. If you aren't a LOPSA member yet, I urge you to consider joining. If you'll read my statement, you know that LOPSA is going to work on behalf of our profession, not just its members. By joining LOPSA now, you get to help that effort by volunteering your time to do really valuable work that will help a lot of people.

As always, if you have any questions or concerns, please comment below or send me an email at standalone.sysadmin@gmail.com.

Thank you,

Matt Simmons
Candidate for the 2013 LOPSA Board of Directors

If you can't see the embedded video, you can watch it at YouTube.

Also, the rest of the talks are in a playlist, so watch them all!

Here's the rest of the day's schedule (in Mountain time):

Wednesday, Apr 24  08:00-10:00	Exablox 
Wednesday, Apr 24  10:30-12:30	Cleversafe
Wednesday, Apr 24  13:30-15:30	Marvell
Wednesday, Apr 24  16:00-18:00	Cisco

The funniest part of the whole situation was that even if there were absolutely no changes, the sync still took 15 minutes. 15 minutes to not sync data? How? Well, because in order to know that you shouldn't sync data, you have to know what the files look like on each side, so basically, rsync inspected the metadata of each set of a half million files on each end, and sent the (compressed) results from one side to the other, then compared the two to see which files should get copied.

I had this much trouble with a half million files….you've got to wonder how companies with actual Big Data problems deal with it. More and more, they're moving to object-based storage solutions. That's what Cleversafe is offering, and as you can see in their case studies, there are some really big datasets living on their hardware.

The Cleversafe offering is actually presented through a series of three separate pieces of hardware, which function as specific parts of an involved cluster solution. The hardware is:

The dsNet Manager (which comes in two flavors: 2100 (offering the ability to manage up to 10PB per dsNet Manager) or the more hefty 3100 (which offers a whopping 100PB of storage under management).

The second piece of the puzzle is the Accesser (something that my spellcheck absolutely refuses to recognize). It's available in two models, as well: the 2100 (which can provide up to 75MB/s of throughput) and the 3100 (which gives an order of magnitude more throughput, at 750MB/s).

The final piece of the puzzle is the actual storage array that holds the data. These guys are the SliceStors, and the two models are 2210 (which holds 36TB raw) and 1440 (which holds 144 TB raw).

Object stores, how do they work? Kind of like portals. Speedy thing goes in, speedy thing comes out.

You don't (usually) address object stores by using anything so barbaric as a filesystem. You use an API, and in the case of Cleversafe, you use a REST API (which looks a whole lot like HTTP's GET, PUT, DELETE, and LIST commands. You $VERB an object in software (so if your software deals with objects like, say, files, you PUT the file into the object store with the API). You are returned an Object ID, which is a unique identifier, or a key that you can use to pull (or GET) the object that you stored.

On a (much) lower level, these devices are using Reed-Solomon Erasure Codes (that link goes to a PDF, and fair warning, you only really need the first dozen slides or so). Essentially, you have the need to store data across several data stores, so that even if you can't access some of the data stores, you can get your data back. This should sound very familiar if you know how RAID 6 works.

Well, as it turns out, you can do some matrix math with the data you want to store and the number of copies you need, and if you have n blocks of data, you can generate m "coding blocks", and distribute n+m blocks throughout the various storage arrays. In order to rebuild the object that you just dispersed, you just need to return n+a, where a is smaller than m - ANY blocks, as long as you've got that number required. Think about that for a second.

What you end up with is a very interesting solution, where the fastest blocks that get returned to you are always the right answer (because the answer can be derived from the contents of the blocks you got - always). If all the data is local, then you get all of the data quickly. If all of the data WAS local, but one of your local arrays blew up, and now half of the blocks of data are local and half are in Zimbabwe, then you still get the data, just not as quickly as you would have.

Erasure codes are definitely what life after RAID looks like, and it's great to see more solutions coming online built with this technique. In terms of the clustering, the SliceStors use Hadoop and MapReduce, technologies built to provide wide-spread data access and analytics. I've never had the chance to play with either of them, so I'm not as familiar as I'd like, so I'll definitely be paying attention when Cleversafe is talking about their architecture and capabilities.

Cleversafe will be the first presenter at Storage Field Day 3 in Denver, on Wednesday, April 24th, starting at 10:30am Mountain Time. You can watch live at Tech Field Day.

The overall purpose of this legislation is actually decent. The idea is that cybersecurity threats are more easily defensible if people share information about attacks. This is exactly the same idea as the RBL anti-spam lists that tons of people subscribe to. The problem is in the details, though.

The text of the bill isn't actually that long, but one of the things that you'll see on Page 9, is Section 1104b(4) is "Exemption from Liability". Here's the text:

1 (4) EXEMPTION FROM LIABILITY.—No civil or
2 criminal cause of action shall lie or be maintained in
3 Federal or State court against a protected entity,
4 self-protected entity, cybersecurity provider, or an
5 officer, employee, or agent of a protected entity, self-
6 protected entity, or cybersecurity provider, acting in
7 good faith—
8 (A) for using cybersecurity systems to
9 identify or obtain cyber threat information or
10 for sharing such information in accordance with
11 this section; or
12 ‘‘(B) for decisions made based on cyber
13 threat information identified, obtained, or
14 shared under this section.

So essentially, any efforts made to investigate attempted breaches are exempted from criminal or civil cases, so long as the efforts were made in good faith? Really? What kind of aggressive action does that exemption extend to? If I see attempted breaches of my web server, do I have free reign to trace the visitor back to the source and try to breach their security in return, so that I can try to determine if this is a bot attack or a concerted effort?

Under this provision, it seems like I do, and that I can't even be taken to civil court for damages caused during my "investigation". Does that seem right to you? It doesn't to me, either.

CISPA has passed in the US House of Representatives. Demand Progress has a campaign to ensure it dies in the Senate. If, after reading through the bill, you also agree that this isn't something that you want passed, then you should contact your Senator. If you aren't in the United States, this law still affects you, since you can't bring a lawsuit in the US against any entity that hacks you back...you know, in "good faith". The EFF has a page for non-US folks where you can sign a petition.

I'm not telling you what to think. I'm only telling you what I think. And I think this sucks.

Modern storage solutions seem to follow one of two mindsets. Either you believe that flash is the current best-of-breed technology for storing data and that you should use it exclusively, or you believe that there is still a place for spinning disks, and that flash should be limited to what it's good at (which is generally to speed up random IO).

While there are a lot of storage companies doing amazing things that follow the former model, I tend to take the viewpoint that while flash might be the fastest way to deal with storage IO, there are still, as of 2013, drawbacks that make a hybrid solution, like that of Starboard Storage, much more palatable to me.

To give you a brief overview of their line, Starboard Storage builds the Application Crafted (AC) line, including AC45, AC72 (basically starter models), and the big boys, AC200, AC4000, and AC4500. Here are the stats breakdown per node:

You can get more details on the "tech specs" PDF, but essentially, the hardware is not the magic for this kind of box. It wouldn't take a lot of creativity to go to SuperMicro and put together a piece of hardware that looks a lot like this. What really makes a difference for a solution in this space are the capabilities and manageability of the box itself.

The Starboard Storage devices run the not-so-creatively-named, "Starboard OS", and it manages the interface between the physical storage devices and the somewhat-more-creatively-named MAST, or Multi-protocol Application STorage. This set of stacked layers is what provides the very cool features that make the Starboard solution so interesting.

Historically, a bunch of drives were RAIDed together. That worked, but it wasn't very manageable. The RAID sets were the only layer of abstraction, and were of a pre-determined size, and not generally fun to resize or carve up. To remedy that, storage pools were added - layers of abstraction on top of RAID groups which could be carved up into pieces, and those pieces presented. The pool was usually tied to a specific RAID group, but it at least offered flexibility that wasn't present before (and the RAID groups at the bottom could be arranged onto specific speeds of disk so that you had some degree of performance choice in the storage pools above).

What Starboard offers is essentially a wide variety of RAID groups at the very bottom, and they refer to these as "Layouts" - they include drives devoted to data, parity, and spares. The sizes are dependent on the number of disks and the capacity of the disks, and performance is mostly related to the number of disks and the performance profile of the disks contained in that particular layout.

Interestingly enough, the same physical disk can belong to multiple layouts at the same time. If a disk belonging to multiple layouts fails, each of the layouts rebuilds its own spare - so assuming the same disk isn't a spare in multiple affected layouts, then the writes required to rebuild get spread across multiple disks, and it doesn't take a week to recover from a drive failure.

The head unit has 16 built-in drives, some of which are taken up by SSDs (the number of which varies depending on the model you get), but each port has a SAS2 6Gb/s interface. There are also expansions available through SAS2 interfaces, with additional shelves having either 16 or 48 drives.

At the top of the stack, data access is made available over a large array of protocols - from file level SMB/NFS to block level iSCSI and Fibre Channel (FCoE? - I didn't see an HBA).

The most interesting part to me is the way that the Starboards handle random IO. I'm familiar with the Nimble Storage method, since they've also appeared at Tech Field Day (and I have since bought one), and the Starboards handle random IO in a similar, but more sophisticated method.

Both the Nimble and Starboard device examine stream IO profiles, and use their SSDs as an acceleration on random reads, but the Starboard devices use the SSD layer as a write-back cache as well, to increase write acknowledgement, then copy the data to the spinning disks underneath at the array's leisure.

At first blush, it might seem blindingly obvious that this is what you should do, but if you dig a little deeper, it's not so clear. SSDs aren't a magic box that work the same way as spinning disks, just faster. Their IO profiles are completely different because of the way that they need to store and retrieve data. Reads are almost always very fast - there's no seek time, so random reads are almost identical to linear reads in terms of latency. Writes, on the other hand, are quite variable depending on a lot of things - the current used capacity of the disk, whether garbage collection is running (or even enabled), whether the filesystem has enabled TRIM support, and so on.

The folks at Nimble took the unquestionably safer stance that since random reads are the biggest win, they would concentrate on providing fast random reads through SSD. The Starboard folks are being much more aggressive by piping random writes to the SSDs, too. I'm curious to hear how much tracking they're doing of the various S.M.A.R.T. fields on the SSDs, whether they're enabling TRIM, whether they're proactively replacing disks before write amplification starts to affect ACK speed, and some other things that I'm going to ask about, and will include in the review I do following the event.

I'm really looking forward to seeing them present on Thursday, April 25th at 10:30am Mountain Time. You can tune in live and watch, too!

Windows 7 introduced the Network Level Authentication option for remote desktop. The two options for remote desktop are:

• Allow connections from computers running any version of Remote Desktop (less secure)
• Allow connections only from computers running Remote Desktop with Network Level Authentication

Using NLA is actually more secure - it used to be that a session would get created, then the user connecting would get authenticated. That led to the potential for denial of service by exhausting the remote machine's facilities. NLA is basically a way of saying, "Make sure the user is who they say they are before I build everything needed for a remote desktop session".

Unfortunately, the tried and true rdesktop doesn't support NLA, which means that if your domain or computer administrator has enabled NLA authentication, then you're out of luck.

Fortunately, there is a fork of rdesktop called freeRDP which includes support for Remote Desktop with NLA. Chances are good that it's in your distro's repo, so check for it.

I've found that the command line arguments are a little bit...umm..fussy. Like, some arguments need to be in a certain order and the arguments aren't in the format that you might expect. Here's what finally made my connection work:

xfreerdp -u msimmons@mydomain --sec nla -g 1280x1024 remotemachine.fqdn

The "--sec nla" appears to be the default, with the other options being 'tls' and 'rdp'. It's very important that no arguments come after the hostname you're connecting to, and the 'mydomain' part is either the domain or the FQDN of the domain. The "normal" username formatting of "DOMAIN\username" doesn't appear to work at all.

So there you go. And there I go, the next time I forget how to do this.

All in all, there were around 15 people at the meeting, and nine lightning talks, covering a wide range of topics. The talks were recorded, and we'll be releasing them individually in the near future.

Here's an overview:

Matt Finnigan gave a talk discussing the LOPSA Mentorship program. If you aren't familiar, the mentorship program is a free service offered by LOPSA, where any admin who needs help, either with a project or just general career guidance, can sign up to be connected to someone with experience in their target area. You need to be a LOPSA member in order to be a mentor, but being a protege is open to anyone, regardless of LOPSA membership.

Adam Moskowitz gave a talk discussing cooking for system administrators. He appealed to our sense of making things as well as our need of healthy food and good value. Adam encouraged us to try cooking, and although most people thought it was expensive to property outfit a kitchen, he reminded us that it was actually a fraction of the price of our new laptops, and the kitchen gear would last a lot longer.

K M Peterson gave two talks, both about interesting technologies that I wasn't aware of. The first was a result of his search for a provider-agnostic method to send SMS messages that didn't break the bank or involve maintaining an array of modems. He ended up setting up a script to talk to Amazon's SMS service, and provided us example code in his slides.

His second talk was on Shamir's Secret Sharing Scheme, aka 'SSSS'. The idea behind this crypto tech is that you have a secret which you want to ensure can only be recovered by the collaboration of a minimum number of involved people - say three of your team of five. You encrypt the plaintext and generate as many keys as people you have, and tell the app how many should be required to release the information. To pull the data out, you provide any of the generated keys, as long as the number of different keys meets the minimum determined when the data was encrypted. Really fun, fascinating stuff. Incidentally, the Shamir is the 'S' in 'RSA' . Cool!

Nahum Shalman gave a really nice introduction to SmartOS, a derivative of OpenSolaris which is maintained by Joyent. Interestingly, the Linux-native KVM was ported to the SmartOS kernel, allowing creative and secure uses of jails and virtual sandboxes, all taking advantage of native ZFS, dtrace, and all kinds of delicious Solaris-y goodness.

Sheeri Cabral came from Mozilla to talk with us about how they're deploying MySQL using Puppet. Her slides had example code, and she walked us through the abstracted object and up to the deployment on the actual nodes. Since we're rolling out a relatively extensive puppet infrastructure at work, I paid close attention, and I'll be following up with her to ask more questions!

John Jarvis talked to us about a creative use for his Raspberry Pi - he securely erases flash media using Stick Destroyer. He rigged up a light so that you have a nice visual indicator of when the stick is being erased, and when it's done. Honestly, it looks really cool, and definitely the first hardware hack I've seen at a sysadmin lightning talk. Nice.

Pat Cable showed up to talk about Sensu, a ruby-based monitoring solution that uses AMQP queues to distribute tasks around a monitoring infrastructure that can scale out horizontally to monitor extremely large numbers of machines. It's definitely a "next gen" monitoring solution that you should be aware of.

Lastly, I got up in front of everyone and talked briefly about something that I've noticed - mainly about how I see our profession splintering, but that the splintered elements (such as network and storage administrators) aren't actually specialties of "system administrators", it's much more like the specialized administrators are specialist doctors, and system administrators are like general practitioners. The idea is still half baked, but that's the fun of a lightning talk, right? I didn't offer any answers, but I asked a lot of questions.

Anyway, as I said, the individual videos will be up before too long. Many thanks to Sheeri, who recorded the sessions for us and will be doing the heavy lifting of editing them apart. I'm very grateful that she came and spoke as well as recorded the session AND brought us CDs of the 100th episode of her podcast, OurSQL, featuring an interview with Randall Munroe, of XKCD fame.

We all had such a great time that we're going to look into doing something similar again in six months. The next community-sourced meeting will probably be a "show and tell", where you can come and talk about whatever cool (or not so cool) things you've been up to. Watch here for more info!

All told, 235 people responded, and that's a pretty good sample size. I think we've got enough to go on and draw some inferences. Before we start, though, you should know - this survey is a prime example of selection bias. I didn't talk to ALL system administrators. I talked to sysadmins who read my blog, or go to the same places online that I go. It isn't necessarily indicative of sysadmins in general - just sysadmins that hang out where I do.

Lets talk about the standardized answers, first.

First, out of curiosity, I wondered if the people answering the survey were, in fact, system/it/whatever administrators. 8% of people responding were, in fact, not currently employed as a sysadmin. I'm guessing that most of these folks are out of work, but there are probably some aren't IT Admins. I would guess managers and the like, but if you read my blog and you're not related to technology, please comment and let me know! And thanks for reading!

One of the more off-the-wall questions that I asked was, "In what month were you born?". A friend of mine and I had a conversation where he observed that a lot of system administrators he knew were born in and around November. I don't believe in horoscopes or anything, but I can see how spending your first impressionable months outside in the sunshine versus inside away from the cold might affect people in broad strokes. So I figured I'd ask. As it turns out, there isn't any kind of statistically important distinction here:

Next, I asked people how content they were with their infrastructure. No big shock- the vast majority were happy with some of it, but not all. I am a little surprised that a full 20% of people are "very content", but I'm happy for them :-)

The "How has your salary changed in the past two years" is interesting. Half of the people had it increase "moderately", around a quarter stayed the same, and around a quarter increased dramatically. Only a thin sliver decreased at all, so that's good news on the economic front.

I asked about LOPSA, because I'm interested in understanding how it's perceived by members of the profession. More than half didn't know about it, and there were some people who weren't interested, and 19% of the folks responding were current members.

Since I also like conferences, I was curious who had been to which conferences. No surprise that LISA is winning in terms of the named conferences, with LOPSA-East and Cascadia listed pat that. There were a LOT of write-ins, too.

When I was asking people if they had any burning questions they wanted to ask system administrators in general, one of my buddies was curious about how people used their monitoring systems, and whether or not they used 3rd party dashboards to organize and present information through a "single pane of glass" type interface. Here's how people answered this:

Multiple people that I talked to were curious how people saw system administration - whether they considered it a career or just a job, and whether it was even what they wanted to be doing. Of the responses, 81% of people consider it to be a real career (although a statistically significant number of people indicated "other", then commented that it was a lifestyle. Take from that what you will.)

Someone I spoke with was interested in how IT Administrators communicated with their teammates and coworkers, so I added that question. Email was the decided winner, although a large number of people run internal, private IM services, and some people use public IM, as well.

And lastly, I asked people about whether they were using a "cloud" provider (or planning on it). I limited it to PaaS and IaaS since I don't really consider SaaS to be a "cloud" service. If so, I've had my email "in the cloud" since 1996. I don't think it counts. Anyway, I'm digressing. Sorry.

OK, so those are the "standardized" results. And only of mild interest, I assure you, compared to the FULL results in the spreadsheet.

Reading the paragraph responses is always an enlightening experience. I love hearing how other people see the world, and what they're doing in their environments. It really opens your eyes to new experiences in a way you don't get otherwise.

Here are some more interesting comments from the results:

The future is OpEx over CapEx

We are a cloud provider ... and I might be more comfortable with the competition!

There is a TON of stuff to read in there. I've pulled out as much personally identifying information as I could find, and made the rest public.

Please pick up your copy here on Google docs:

https://docs.google.com/spreadsheet/ccc?key=0ArmQ8DpzeTPbdFNBRUJZYkR6TWl6MVNtTEk0YkdHN2c&usp=sharing.

Thanks, and please, leave any feedback you have in the comments below!

I also need people to sign up to present! Yes, this means you! Fill out this form and let me know what you want to talk about:

It's only 5 minutes (or less), and it's fun, so come and tell people what you currently love in IT!

On one extreme is VMworld, epic in scale, which takes up a sizable portion of the convention space in whatever city they decide to locate it in.

At the other end of the spectrum are small conferences like Monitorama. Accurately described as a "single track", it's two days of high signal, low noise immersion among the DevOps practitioners and elite, with the entire second day devoted to workshops and a "hackathon", where basically you sit around hacking on whatever cool technologies you want, with whoever you want. It's awesome.

Today, we had two keynotes, one by John Allspaw (@Allspaw), Senior Vice-President at Etsy, and Dr Neil Gunther, (@DrQz).

Allspaw's keynote discussed monitoring considerations, including the reminder that we really need to learn from other disciplines. One of his key messages was that other professional domains have a common language or lexicon, and if we're going to interoperate with them and truly learn, we need to adoptt their shared language. We also need to establish a relatively standardized language among ourselves.

I really enjoy hearing John speak, and this message is right on target with what he's been saying, but one thing he mentioned that I hadn't considered before was that we actually have some advantages over other industries.

We might have relatively unsophisticated analytics and practices, but not only can we learn from other industries, we are going through our growing pains in the information age - we literally have the world's accumulated knowledge at our fingertips.

Dr Gunther's keynote was also very interesting, and he attacked monitoring from the perspective that all data is, by definition, incorrect. The very fact that we're rendering reality in the form of discrete numerical representations means that we're dealing with approximations. The real questions are:

1. What is the degree of error?
2. Is that degree of error acceptable?

Overall, there were 12 sessions and one panel, but in terms of content, it was absolutely packed. I seriously had a great time. It seemed like there was a lot of sitting, since only having one track means staying in one room, but there were several breaks and lunch to break it up and provide the all-important hallway track.

I don't know if Jason Dixon is planning on having it in Boston again next year, but I don't think it matters where it's held. I'm there, because the was the most valuable $100 I've spent in a long time. Even if the talks weren't packed with technical howtos, it was awesome to be exposed to so many new ideas and surrounded by so many people who had their stuff together. I would highly recommend this conference to anyone interested in web or devops or monitoring in general.

The leather stuff I'll be giving away is:

• 5 leather keychain fobs, one in each color they make (yours will be chosen at random)
• 1 leather cup holder to keep hot coffee from burning your hand
• 1 bottle of car Leather Care
• 1 bottle of furniture Leather Care
• 1 ID wallet (the other side of the wallet has a window for your ID)
• 1 iPad 2/3 case

This is all Saddleback Leather, some of the best stuff you can buy. The iPad case by itself is over $100!

The iPad case was going to be the biggest win, but now there's something even better. Courtesy of Scott Murphy of Ottawa Valley SAGE and LOPSA, the grand prize is FREE ADMISSION TO LOPSA-East! How awesome is that?!?

"But Matt! How can I win such great and amazing prizes?" you might well ask.

Take this survey!, I would reply.

Seriously, I've put together a survey of...frankly...assorted questions, from myself and other people that I'm friends with, all of who are curious about different aspects of life as system administrators. Please take some time and think about your answers.

And if you're curious, I will be releasing the results of the survey publicly. As always, I will redact any personal or identifying information, and if I have questions, I'll be in touch (since one of the fields in this survey is for your email address, that way I can let you know that you win).

If you can't see that embedded iframe, you can click here to take the survey.

Today is Friday, March 22nd. I'm going to be shutting down this survey on Friday, March 29th at noon. I'll spend the weekend redacting personal details and starting to contact the winners, and I'll release the survey results early the week of April 1st.

Thanks for your time, thanks for filling out the survey, and if you have any questions, please don't hesitate to reach out to me here or via email.

Read about them here!