Marcelo Calbucci

    If you have a Visa to work on the US, you are probably on a tough spot. Davis Bae, at the Bae Law Group is offering a free presentations at their office in Seattle as to what are your options.

    Please, foward this to any of your friends that might need this information.

    More info at their website:

    http://www.baelawgroup.com/articles_story.asp?tdx=1&ndx=27

    About 4 years ago I wrote a blog post asking why every new blogger had to write a blog post about blogging. I guess Twitter's time has come. Every new Twitter (or old Twitter) has to write about how to use Twitter, how to gather more followers, what tools to use, yada, yada, yada.

    All that I want to give you today are two very practical tips:

#1 You cannot send a DM (direct message) to someone that is not following you!

   How many times you saw a tweet like this: "Send me a DM if you want to talk about X". Turns out that you this person didn't get any DM, most likely. Why? Because people cannot send direct message to you unless you are following them.

#2 Only Mutual Followers Can See @Replies

   This is a bit harder to explain, but by default, if you send a message that starts with a @reply as in "@calbucci Marcelo is a great guy, you should really follow him" only the people that are following you and following that person (on this case @calbucci) will see this message. For example, if you re-tweet a message (aka RT), but you re-tweet w/ the sender Twitter account as the first word, only the people that are already following that person will get it.

   I'm not sure I'm being very clear on what that means, but so you understand a bit better, go to Twitter -> Settings -> Notices and see the @ replies options. By default, you'll only see replies to the people you are following.

    And just to be extra clear, @replies will appear on your tweet timeline for everyone to see, but they won't appear on your followers "view" unless they are also following that person.

    Ok, the title is a bit misleading, but the point does relate to it.

    I have a great friend that sends some emails that are a bit large and sprinkled with questions or points for you to give input. Do you think people answer his email? Yes, they do, but they answer only half the questions or just 1 question (usually, the last one).

    The problem when you ask someone too many questions, or ask them to do too much for you (send me this, then that, then go there, bring this) it's very hard for them to know what they should get done first. Or, which thing is "the" thing that must be really accomplished.

    The other day I need two things from a person. I need an urgent headshot and I need a short bio. I sent the email to this person, and being direct I asked "Can you send me a picture of you and a short bio?". That was it.

    Immediately I realized my mistake. He's a very busy entrepreneur. He would read that email and just file it under the pile of "must do". If I had been smarter I'd have asked him for a picture and he certainly would have one at hand (everyone has a headshot ready this days to quickly upload to the next social network you join). After I've got that, which was the most important piece I needed, I would have asked him for a short bio.

    I'm getting a lot of people to agree to a lot of things lately and I think I'm getting that by slowly bringing them into the "full ask" without scaring them away with the "time hog" thinking.

    So, the next time you send an email to someone to ask for something, start by asking just 1 thing that can be easily accomplished and reel them in.

    I just came back from the Microsoft Entrepreneur Advisory Council, a group of local entrepreneurs put up by Microsoft Emerging Business Team to talk about what startups need. I felt humbled by the presence of such great entrepreneurs on the table, me probably being the worst of the bunch.

    This is very smart from Microsoft for two reasons: Group dynamic always reveals interesting arguments in favor or against something, and, second, startups are where the future lies.

    Not long ago Amazon invited me to the Amazon Start-Up Challenge Award in 2007 (they forgot to invite me for 2008). And I said the following phrase which they event quoted on the Homepage of the Amazon Web Services site:

"...Get the CEO/CTO of small startups and get them engaged and talking. A few of those will become CEO/CTO of Fortune 500 companies and all built on top of Amazon Web Services."

    Who would have thought that Microsoft reads my blog and got the message?

    One thing became crystal clear on the lunch today: Microsoft is *very* disconnected from the Startup world. The praise and talk about their solution in a way that is un-appealing for Startups. On the other hand, all that rhetoric would work wonderfully with the CIO of a Fortune 500 company.

    I think they get that. They get they are the "under dog" on this new web + startup world and that's why they were so interested in what we had to say.

    They talked about Silverlight and Azure (which doesn't pronounces the way you think it does). Silverlight is easy to understand because you can simply compare it to Flash, in other words "Silverlight is Microsoft's Flash". Azure on the other hand is not "Microsoft's EC2". I've been following Azure from the sidelines (side note: Mohit Srivastava, founder of Faves, just moved back to Microsoft to work on that project).

    I'm very impressed w/ Azure. Although it doesn't seem ready to prime time, it will be shortly and I'm having serious thoughts about moving Sampa entirely into their platform. It's a great response to Amazon EC2, but done in a more "interesting" way. It does seem to be a real "Cloud OS", where the Sun's dream finally comes true ("the network is the computer").

    I don't think Amazon has much to fear from Azure since they are competing for a very different crowd, one being the LAMP/Ruby guys, the other the .NET developers.

    Silverlight on the other hand has bigger problems. The decision of moving Sampa to Azure is totally under our control. If we think is cost-effective, we do it. The decision to create any part of Sampa in Silverlight is mostly outside of my control. User adoption is the barrier. Consumers get very scared of installing any piece of software on their computer from a website. Even if 90% of people didn't care and install it, it would mean 10% would simply turn back and go away. We can't afford that unless the benefits (read, customer retention) outweighs the fear of installing Silverlight plug-in.

    Now, I have a final piece of advice to Microsoft: Don't look for inspiration for innovation on the enterprise. Innovation comes from Startups. Although the enterprise revenue probably is 100x the startup revenue for you, every piece of innovation in computer history came first from a startup (or a developer/researcher thinking like a startup inside a big company). So, stop asking Boeing, US Gov, BofA or GE what they want and start asking BuddyTV, Redfin, Smilebox, Picnik, RescueTime what would they like in terms of development tools, platforms, servers, etc.

    Stay tuned for a blog post about Silverlight and the future of the Web.

    I woke up this morning and saw some unusual events on my inbox. Sampa servers have lots of self-monitoring and self-healing, but when things are not well I get notified. This morning, our server friendly named SADCWS01 was not responding.

    This was our first server. The very first server I bought for Sampa almost 4 years ago. Every once in a while, I have a server that reboots itself to try to address some memory or connectivity issue and the server doesn't come back by itself. It actually happens once every 3 months or so.

    At 8:00 AM this morning is when I saw the issue. The first order was to try to reboot it remotely, but that didn't work. I could have asked the datacenter to reboot the server but I had a feeling things were not right, so I drove to our datacenter, attached the monitor to the server and rebooted it manually.

    Oh-oh! The boot screen says something ugly, like "your hard disk is not operating under normal parameters, we recommend you backup it...". Gosh, it happened.

    The day every entrepreneur, or every ops personnel, or every IT person fears. We have a loss of a hard-disk.

    I absolutely knew this day would come, and I design the Sampa software and architecture to survive this, but I knew it would take time to bring everything back online.

    This was about 9:45 AM and I was sitting in the floor on the datacenter, very cold and very noise, pondering my options. Do I take the server to the office, buy a new HD, replace it, restore the data. Do I take the good HD out (each server has 2 HDs that backup to each other) and restore on another server? Do I try to kick the server a bit (maybe it would magically come back online)?

    Well, I decided I didn't want to "save" that server. Once a server presents a problem like this, you better just get rid of it and move on. You never know if the HD issue was because of the HD, or a bad controller, bad cable, power supply surge, etc.

    So, I went to Hard-Drives Northwest, just a few blocks from the datacenter, bought an SATA Enclosure w/ a USB connector and went back to the datacenter.

    The first issue was this server was at the bottom of a stack of servers (5). I couldn't remove this server without re-arranging all other servers, so I shutdown all the other servers as to not cause another HD issue on another server by some clumsy act. Then, I carefully started to remove this server from the stack, while pushing the other servers on top of that one back. It probably took me more than a minute to do that very slowly, always checking no cables were getting squished. At the last inch I pushed a bit too much and all the servers fell hard on the shelf. Crap! It was just an 1in drop, but still.

    I open the "dead server", remove the hard disk, put it on the enclosure so it becomes an external HD and connect to another server. Go check the screen and YES! I have the backup data and it seems good. Let's start the restore.

    By now this is 11:00 AM. I was being very meticously about it as to not screw up anything. We are talking about 10,000 customers here. Their blog posts, their pictures, the baby milestones. My own blog (the one you are reading right now) was on that server, but worse of all, my wife's blog that she writes about our kids are on that server.

    After 15 minutes of restoring the data I'm able to calculate how long it's going to take. Nothing short of 6 hours to copy everything to a new server (about 200GB of data) which means copying will be done by 5 PM today.

    As of this writing, 65% is completed. Once I restore the data into the new server I still need to make the DNS changes to make sure the sites point to the new server. Good thing our system expires DNS entries in 1h, and not the default 72h of most sites.

    I can't say that I'm happy with the way things turned out today, but I feel relieved I was able to restore the data, but about 10% of our customers are not able to access their website today and they don't even know why (our blog is also on that server).