Franco Dal Molin Glimpsing into the Future of Collanos

As founder and CTO, I would like to initiate a dialogue on our company blog about the technical aspects of Collanos’ offering. This article discusses the main limitations of today’s Workplace release as well as our ongoing activities to address and rapidly fix the problems. The focus is on fundamental platform evolution. I will not talk about business features or collaboration functions today. In my next blog postings, I will shed more light on other aspects like Collanos’ vision, product portfolio and feature roadmaps.

Collanos had been in “beta” for about half a year before dropping the “beta” label two months ago. We have come a long way, but the journey has just started. There are many exciting new features we plan to release to our growing user base. This post introduces some of the features we will release in the coming months.

Starting Out
In eight months, we went through a dozen of small releases, fixing nearly 400 bugs and issues. Don’t be surprised; this is software engineering best practice: quick turn around cycles and “perpetual releases”. We also added many features, and with every improved version, the number of downloads continued to increase. We are about to reach 40,000 downloads, and it is exciting for us to see that famous “exponential curve” (aka the ‘hockey stick curve’) kicking in.

Invaluable Feedback
A big thank you goes to our active user community. Your great feedback helps us better understand what works, what doesn’t, what you like and dislike, and what you miss. This blog post is a summary of the main limitations of our product and how we are addressing them. If you are interested in raving quotes and why users are enthusiastic about Collanos, you can find them on our web site. We have about 30 rotating quotes on the Products page, and new quotes are added daily – just refresh the web page regularly to see them.

Main Limitations
While we receive great feature requests and lots of ideas from users, many could be classified as “nice to have” or are subjective to a very specific use-case. Still, there are a number of issues that come up frequently. In fact, there are important limitations in our current product release that prevent some of our users to deploy the solution more widely. Here is the ranking:

  1. User must be online to be invited – An invitee must be online, otherwise there is no way to invite this person. This limitation comes from our current “pure” P2P model with no additional centralized servers or “helper” services.
  2. Conflicts can lead to data loss – If two users make changes to content at the same time, then one of the two will likely lose his/her changes. This is a limitation of the current simple “last modifier wins” rule. BTW conflicts cannot be avoided in a P2P system – only mitigated and managed.
  3. All content must be received from its originator – The current P2P replication algorithm fetches data from the source (i.e. its originator) only. If the originator is currently offline, its new content or latest changes cannot be received even if another team member is online with that new content.

Near Future
We are addressing all of the above issues with high priority. Our goal is to relieve “user pain” with as simple and pragmatic solutions as possible. We could certainly think of very elaborate and complex “enterprise class” feature sets, but this would only delay the process. Our mantra in engineering is: Deliver great user value fast. In the near future we are going to ship the following features (in this order):

1. Central User Directory – This new central service serves multiple purposes. First of all, it’s a directory for our community so users can easily find other users regardless if they are online or not. You will be able to search by different attributes – much like current social networking sites. And no matter if a person is currently online or offline, you will be able to invite this person.

2. Conflict Bin – This simple “save my data” concept will prevent users from losing their work due to unexpected conflicts. If a conflict happens (again, this is unavoidable in P2P networks), then the overwritten (or even deleted) copy will be saved into the Conflict Bin of the affected user. From there, the data can easily be recovered. At a later stage, we might expand the concept and add support for file versioning. With this, users will be able to refer back to any previous document version.

What’s better? These two features will be rolled out within 3-5 weeks so stay tuned!

3. Improved Replication – Another major effort is going into bringing our replication algorithms to the next level. We are building in some advanced capabilities and intelligence like a peer’s ability of fetching new workspace content from any other peer who already received it, or fetching it from many peers at the same time. Other measures will include advanced optimizations such as measuring P2P bandwidths or dynamically applying priorities. It will also be possible to see how much data is left for transmission on a per object basis.

As you can imagine, the third feature set is a bit more complex to implement. We plan to roll out features gradually with the first set of improvements due in approximately two months. This second-generation synchronization will be the single most important improvement of the Collanos platform. The benefits for our users will be substantial. Users can expect faster replication times and higher transparency of what is being replicated.

And Beyond…
We have a bolder vision, and some of the building blocks are already on the drawing boards:

  • Permanent Peer – This service will be an on-demand subscription offered through ISPs. It will boost your overall Collanos experience. Think of it as your other always-on peer that not only maximizes up-to-date workspace content, but offers backup, workspace management, and more.
  • Web Workplace – This extension will come with the Permanent Peer. It essentially brings the Workplace to your standard web browser, allowing you to access your workspaces from any connected device. This flexible hybrid architecture will make Collanos the first seamless Peer-to-Peer-to-Web offering.
  • Voice Services and Instant Messaging – We will integrate VoIP and more general instant messaging services with our Workplace very soon. The idea is to facilitate individual and team calls directly from the context of your various workspaces. Imagine a one-click conference call with your team!
  • Multiple Languages – With the help of our user community, we recently added German to the Workplace and will soon add Chinese. More languages will be added this year. If you are interested in contributing with a specific translation, please get in touch with us.

I hope this information was valuable for you. Do you think we are addressing all the right issues? Do you have questions or would you like to see other topics covered? Please let me know what you think.

Sphere: Related Content

You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

Share with Shareomatic! Share with Shareomatic! | View blog reactions

10 Responses to “Glimpsing into the Future of Collanos”

  1. Peter Helfenstein Says:

    Todays PhillyTechcast with Collanos user Steven Pearl confirms the importance of the improved replication for all our users. This blog series will address Steven’s requirements re IM and calendar soon!
    Interesting webcast:

    http://www.phillytechcast.com/2007/03/interview_steve.html

    Such feedback is a great motivation for all the Collanos team!

  2. Steve Pearl Says:

    Here’s hoping that my comments on the TechCast get Collanos the positive recognition it deserves. You guys are doing a great service to the IT community in developing this product. Yes, I acknowledge Collanos “early stage” warts in my comments, but I hope the admiration I have for your product comes through in my comments. Keep up the good work!

    Steve

  3. Martin Kuppinger Says:

    Franco,

    like you know I think that you are on the right track and many of the features I have in mind for Collanos and discussed with you are prioritized on your roadmap.
    So I’m looking forward especially for the central user directory and, in consequence, an easy way where I can have one identity with Collanos even when I work on different systems (like my notebook and the PC in my home office).

    Martin

  4. tra Says:

    Franco,

    Great to see the progress and continued improvements, and glad to see that the JXTA technology, the industry leading open-source P2P platform, has enabled you to develop a great product.

    B.

  5. Mark Beaubien Says:

    First off, great job!

    A couple ideas, (which you may have already pondered):
    1. to prevent overwrites, how about a semaphore/lock mechanism on files? Essentially, you check out a file, and if the file is being edited/changed/overwritten a hash code is generated to indicate to another peer that the file’s been modified since it was last checked-in. Source code control systems do this.
    2. For user must be online to be invited, perhaps use a method of sending them a smtp email and when they open it they click on a URL hyperlink (OK that means a web server and that’s not pure P2P) and if necessary, auto-download/install the entire environment (*gasp*). Then they opt-in to the workspace?

    -Mark +1-413-863-0200 X7201

  6. Steve van Niman Says:

    Franco,

    Thank you for the roadmap and general direction we can expect from Collanos in the near term. Your teams willingness to listen to practical improvments have helped improve our collaboration within our team. We now have timeframes on when we can tap into additional functionality to aid in improving performance.

    Thanks,
    Steve

  7. Neil Johnson Says:

    OK, I just linked here from a “Groove” discussion on Slashdot, and took the tour. I like the idea, and the product looks pretty good. I do have a couple of questions though:
    1. How much memory does the application use?
    2. Is there a version that does NOT have all of the communication and files going throuh a server that is not under my control?

  8. Franco Dal Molin Says:

    Thanks everybody for your comments, encouragements, and ideas!

    Mark, 1. The tricky part when it comes to locking/checking-in-out schemes is being a P2P system and furthermore allowing off-line work. While marking content as “changed” is easy, the complexity is pushed to the “conflict resolution” part (and again, because of P2P, remains still hard). Source control systems are server centric. We are working on concepts such as “managed objects” with a “mobile locking” that can be passed on to others. More about this in a future column. 2. Off-line invites. First step comes with central user directory, but yes, your ideas are the next steps: Email invitations/notifications including opt-in to specific workplaces.

    Neil, 1. Workplace currently is a 45MB download, and typically has 90-120MB memory footprint (depending on active workspaces). This numbers are before applying optimizations. We will unbundle the Java Runtime Environment (JRE), which will make it roughly 15MB lighter. A 30MB download is comparable to Skype. We are also working on the memory reduction. Currently we are using about the same as OUTLOOK, but with the replication improvements we hope to also bring down this number. 2. You can read on http://www.jxta.org about the P2P protocols that we use. Traffic is going through a “Relay” only, if no direct TCP connection can be established between any two peers. If that is the case, still, all communication is encrypted from peer to peer.

  9. Benoit Broucke Says:

    Hello every body,
    I follow Collanos since a few time, and as lot of people my feeling is it looks like a great product. I would like to add two comments. First is regarding the data security and especially about the encryption. I understand the answer of Franco, but it stills a problem: no way to control encryption, and no way to control the bounds of the community. In some companies it’s a problem. Or perhaps it’s just a problem of explanation/information. The second point is the concept of data space, work space, repository space … and associated, tools, binary formats… In a compagny data management is more or less organised, versionning, project repository and others. Using several data model and tools generates data replication, copies, and after a while … which version is right ? Just to think about this organisationnal problem. But I’m not sure a solution is profiling today. (The nature of data storing, where it is, and how I can access it should be tool agnosting and defined in a public concept of “electronic data”, with kinds of properties like security).

  10. Al M Says:

    The web interface to Collanos will be make or break. Right now in our team - half of us use Google Docs/Sites exclusively. For Collanos to compete with the elephant in the room, the web experience is critical. I still think offline access to data (which Google does not provide (for now! but Google Gears is coming!) is extremely critical for businesses.
    AM

Leave a Reply