Are hashtags too geeky for Google+? A meditated answer to Loic Le Meur

I was able to join Google+ last week, and found it very interesting for three reasons:

  1. Circles
  2. There are mostly early adopters, and this generates some very interesting conversations
  3. It isn’t blocked at my office yet, so I don’t have to resort to dirty tricks to use it like I have to do with facebook – yet

Is it all good then? Of course not, there are many useful features that could be added. But the one I really miss is tags. I fell in love with tags since I first started using (before the yahoo-less-licious days) and were delighted to see them spreading to more and more uses. It’s no coincidence that I created a website completely devoted to (hash)tags!

Today I found out I’m not alone when I stumbled upon a post on G+ about the use of hashtags on G+ itself. After all, the use of hashtags on Twitter started exactly because there wasn’t a tag feature there, like on G+ now. And, as I discovered later, the same Chris Messina who first proposed their adoption on Twitter did the same for G+ just four days ago.

Not everybody agrees, though. Loic Le Meur commented on that same post where I first read about the debate that “hashtags are geeky and they shouldn’t be added to G+”.

I answered that hashtags are geeky, but simple tags aren’t. But, on second thought, the real answer should have been: Hashtags can’t be “added” to G+; they weren’t even “added” to Twitter, only half-heartedly supported after their use became widespread.

So the real question should be: Should tags be added to G+? I definitely think they should; is there any better way to allow the discovery of interesting conversations? And: Are tags too geeky for G+? Considering that they’re also used on Facebook, I guess they really aren’t.

So please, Google, add our beloved tags to G+; and, while you’re at it, also add a sampling streaming API for public messages so that all sorts of interesting research could be done and, why not, so that I could also add the data from G+ to :)

OpenVZ how-to: Your own free VPS infrastructure in 10 minutes

As I wrote in part 1 of this series of posts, creating your own dedicated Linux “cloud” for free on a root server is easier than you might think. It for sure has been easier for me than I thought before actually doing it. In the first part of this post I’ll explain my choices, but if you want you can just skip to the how-to part.

My first step was to choose a virtualization technology. There are many free open source solutions that allow you to create and manage a virtualization infrastructure on a Linux host; these solutions can broadly be divided into two different approaches:

  • OS Level (aka kernel level) virtualization
  • Full virtualization

To simplify things, we can say that with full virtualization technologies you can have virtual machines that run any operating system that would work on the virtualized hardware, while with kernel level virtualization you can only have containers (which aren’t full virtual machines) that are compatible with your kernel.

In my case I was only interested in Linux virtual servers on a Linux host, so both approaches would have worked. On a performance and resource usage level, kernel virtualization imposes less overhead, but doesn’t guarantee the same level of isolation between the virtual servers. On my own private cloud isolation isn’t so much a concern, so I decided that kernel level virtualization would be my first choice, if it would result easy enough to set up and manage.

OpenVZ is the most used solution for free kernel level virtualization on Linux – as a matter of fact, most low cost VPS solutions offered on the internet are based on it. So I did my research about how to install and manage it. As usual for free open source technologies, there are a plethora of solutions; in the end I chose two possible options as the most promising:

  • Using a full stack solution, either Proxmox or OpenNode
  • Installing OpenvVZ and a web control panel myself

I would have liked to use one of the full stack solutions, both because of the ease of installation they promise, and because with those it would have been easier to also use full virtualization, if needed, at a later time. But both Proxmox and OpenNode offer an ISO installer as their main option, and I couldn’t use that on my hetzner EQ root server; moreover, even if I could, I then would have had to configure the network myself. This is why I decided to install OpenVZ myself on one of the pre-configured Linux OS images that were available for my dedicated server.

The Linux version I know best is Ubuntu Server, as this is what I used on all my own virtual servers until now. The second one is CentOS, which is the default at my day job – but there I don’t do system administration. So I first tried to install OpenVZ on Ubuntu server, and then on CentOS – installing a new OS image in my dedicated server only takes a few minutes, so making different tries was very easy.

From my search on the internet and my own attempts I found out that installing OpenVZ on either of those two OSs isn’t as easy as it could be – or, at least, as easy as I wished it to be, as I found some problems in the repository versions of OpenVZ (these of course could have been fixed at the time of reading this article). From what I read, Debian is the most supported Linux flavor for OpenVZ, so that’s what I tried next.

The recipe

Installing, configuring and using OpenVZ on Debian was really very easy. The following are the detailed steps I used. After writing a draft of this post I tried them and it took less than 10 minutes from start to finish.

  1. Start with a clean, minimal Debian 6.0 installation – I used the pre-built image I found for my hetzner EQ server
  2. Login as root
  3. Install OpenVZ with its management tools and its special kernel, 64-bit version (this is good for both AMD and Intel 64 bit CPUs):
    apt-get install linux-image-openvz-amd64 vzctl vzquota vzdump
  4. Create a symlink to easily find the OpenVZ files:
    ln -s /var/lib/vz /vz
  5. Edit sysctl config
    nano /etc/sysctl.conf

    to make it like this:

    ### Hetzner Online AG installimage
    # sysctl config

    net.ipv4.conf.default.proxy_arp = 0
    kernel.sysrq = 1
    net.ipv4.conf.default.send_redirects = 1
    net.ipv4.conf.all.send_redirects = 0

  6. Use the new options:
    sysctl -p
  7. Restart the server:
  8. Download a pre configured container image for my VPS – I was interested in ubuntu server:
    cd /vz/template/cache

    You can find more images here:

At this point you are ready to use your VPS. But I wanted to make management easier, so I decided to immediately install a web panel to control OpenVZ. My choice was OpenVZ web panel, which you can install with these commands:

cd /root
wget -O – | sh

There you go! Your web panel is already started and you can find it at port 3000 of your server.

To create your first VPS at this point you only need an IP from your provider and use the web panel:

  1. Open http://<yourip>:3000/ in your browser
  2. Login using admin/admin – don’t forget to change the default password!
  3. Click on “Physical Servers->localhost”
  4. Click “Create virtual server”
  5. Input a server ID – use a number over 100, for example 101 – IDs from 1 to 100 are reserved and you’ll have problems if you use one of them
  6. Choose your OS template – you’ll find only ubuntu 10 if you followed my instructions
  7. Choose your server template or leave the default
  8. Type the IP you got from your provider (I’m talking about an IPv4 here, but on a next post I’ll write about using IPv6)
  9. Choose a host name, like “myhostname”
  10. Click on “Additional settings” and choose how much RAM, disk and CPU you want for the new machine
  11. Choose a password and click “create”

That’s it! Your VPS is ready to use. Just connect to it via ssh, and you can use it as you wish, just the same as if you bought it from any VPS provider. You can also clone your new server with minimal effort from the web panel: You just need to select it, click Tools->Clone, type the new IP for the clone and go on. If you don’t have one more public IP from your provider, and you don’t need it, you can use a private one like

If you’re going to do anything serious with your virtual cloud of VPSs, though, there are at least two more things you’ll want to set up: Backups and a firewall.

A backup strategy is really important here, because you’re working on your own physical server now, and if it fails and any data is lost you’re on your own. Moreover, if you are using a low cost dedicated server like mine, you should always keep in mind that they aren’t based on highly redundant, fault tolerant server hardware. There is still software RAID 1 for the disks, but that’s about it for redundancy.

The good news is that backing up a whole VPS is very easy with OpenVZ. There are three different ways to do it:

  • Simple dump – long downtime: The VPS will be stopped while the backup is going on
  • Suspend mode – very short downtime: The system will suspend the VPS and resume it a few seconds later
  • LVM2 snapshot – no downtime: A snapshot is created without suspending the VPS at all

The third method, of course, is the best one. But it requires you to configure your file system in a special way, and for most situations the few seconds of downtime of the suspend mode are acceptable – they are for me. So I went this way.

To backup a running VPS (container) from the command line (or a batch script) with the second method you just type:

vzdump –suspend NNN

where NNN is the ID of the server you want to backup. This will create a file named vzdump-openvz-NNN-YYY_MM_DD-hh_mm_ss.tar in the /vz/dump folder.

By restoring this backup (with vzrestore) you get back your full VPS in the state it was when backed up. You just need to move the backup file to a secure location (preferably after compressing it – with bzip2 you could have a 75% reduction of the file size); In my case I got a 100 GB FTP backup space from hetzner where I move my backups.

To protect your server/s you should also set up a firewall, at least for the host: By gaining control of it, an hacker could gain control to all your VPSs. This isn’t OpenVZ specific though, so I’ll not cover it in this post.


Setting up my own private “virtual cloud” has been much easier than I thought – and much cheaper. The most time consuming part was finding the information on various wikis, blog posts and other sources. With this post I hope to give something back to the community by packaging the updated information in an hopefully easy to use format, even if I’m aware that all this will become outdated at the speed of the internet.