Wednesday, October 31, 2012


One of the things a company needs is a long term strategy and a clear vision on the market and its future.
Having said that I must admit that as long as I was with ALu I was not always able to see the big picture (as I was part ow a some time orwell-esque world inside). Having stepped back a little and looking to my former employer I have to say that they are trying to reshape the telco world as we know it with three ground breaking technologies:

1. Light Radio
Light radio changes the way the RAN will be built by transforming the large bulky towers into ubiquous cubes able to harvest radio signals with less power and in a denser grid. They are one of the most elegant implementation of the SDR concept and here we can see the fine touch of the Bell Labs. They are able tho modulate everything from GSM, 3G, WiMax to LTE as they are in fact remote radio heads driven optically from the network.

2. FP3
In order to accommodate the traffic the LightRadio will sustain due to the increasing HD video, P2P and such a new transport infrastructure has to be put in place. The FP3 network processor is the device that makes possible to create the switching fabric for the the data tide the is about to come. its 400Gbps capacity is intended for the future and can offer a competitive advantage in an all-IP network.

3. CloudBand 
If the two technologies described above are already on the market Cloud band is not yet there. Its aim is to create a completely virtualizable core network. However this is the trickiest and the most sensitive part of the strategy as ALu is not really a software company. Having the core network as PaaS raises lots of unknown points as core QoS, core HA, core network security. I doubt that the telco operators will ever want to put their money making services in somebody's else cloud as there is sensitive data and the core is somehow quite realtime sensitive - wich puts a lot of pressure on the cloud. As for HA I think what the current providers have is far less than the 99.999% the telcos need (think about the AWS outages).

ALu's strategy makes sense but their timing and execution is endangered. If they succeed (as they have the brain mass for creation) they will for sure be disruptive. Until them they have to finance their research on a shrinking margin market with a fierce concurrence from giants as Ericsson and Huawei.

Monday, October 29, 2012

Commit comments on BitBucket

I have been using BitBucket for quite a while and now, with the new interface I discovered much to my shame how important are relevant commit messages.

After some years of Clearcase where the commit message was "." or the id of the bug the new system exposes the changeset's meaning which is great as it forces to think about the relevance of the commit.

Wednesday, October 24, 2012

Art Exhibitions

I have recently went to two art exhibitions in Timisoara. The first one was master's Ioan Tuculescu. I enjoyed it a lot as I have learnt how he shifted from water colors and academic style to a harsher oil on canvas and a crypto-symbolic language. Totems and imagery from the Romanian tapestries are populating his later works.

The second exhibition I went was Ana Ruxandra Ilfoveanu's (the wife of Sorin Ilfoveanu - a big name in Romanian contemporary painting). The works resemble much the one of her husband, they share the same ancient-like drawing technique (although his is more subtle with elements of rupestre pictures)

One interesting fact about her is that her work was associated with the image of Rotemberg winery (wich really produces a good wine). It is interesting how her drawings illustrate the the premium bottles' labels.

Image from

Wednesday, October 17, 2012

Smart Phones killed Intelligent Network

IN was designed to provide logic for the dumb terminals on PSTN or GSM RAN.
As smart phones invaded the market in the latest years most of the functions the initially IN provided moved to the phone or could be achieved over IP links (Is somebody still using MMS anymore?)

IMHO the only remaining use case of IN is in this moment charging for voice and data services.

Sunday, October 14, 2012


Lately I have witnessed a new trend: LinkedIn endorsements. I have been endorsed and I have endorsed myself several people.

The problem with endorsements is that they are flawed by design. It is possible that an surgeon to endorse a software engineer for knowledge on Dante's works. I have seen HR people endorsing project managers for Java skills and because I know them both I can state that the endorsement is not relevant. I have also seen technical endorsements between people who  had not even shared the same hallway for technical talks, not to mention the fact that they never worked on the same team/group/project or even seen each-other's code or design.

I have no silver bullet idea how would it be possible to make endorsements reliable. One way would be to bet on people's honesty, but...everybody lies.
Another way would be to limit the endorsements to the intersection of the skills the two individuals have hence making it harder to endorse  someone blindly as the intersection of the skill sets would limit the interaction to peers.

A finer grained approach  would be to have the skill sets associated to a position and the endorsements to be enabled only for the skills the endorser end the endorsee have in common if both worked in the same company or the were in a common project or in a business relation for the period.

In the meanwhile I am curious how LinkedIn will solve this.