Liquid Telecom Succeeds in providing milkshakes, fails in connectivity

Last year, I wrote this story of pathetic connectivity at the Connected Kenya Summit. The issue was; how can you have a conference talking about connectivity and one of the most touted regional company can not manage internet connectivity?

This year, Liquid Telecom decided to abandon its core business and succeed in the food and beverage department. While the wifi didn’t work from the word go in the morning, Liquid shakes were working very well and some internet-frustrated attendees enjoyed the milkshakes thoroughly.

2014-04-15 17.09.32

I am sure for some people, having no internet connectivity is good for concentration but you know what? we are so hooked that people switched to their mobile hotspots, which means that in a room of 400 people, there were probably 300 hotspots.

This resulted to Liquid Telecom blaming these little hotspots for the failure in their network set up. Apparently these little hotspots were clogging the network. This didn’t make sense.

I have attended the Mobile World Congress, which is attended by 100,000 people and each provider has their network and yet, the public wifi works very well. How does it work?

Anyway, I think Liquid had a problem with their network and for a company of their repute, I think we should have gotten a better excuse/reason.

Apparently, the organisers were promised an STM1 (155mbps) and Ben Roberts, Liquid CEO confirmed that they were pushing 170mbps which is huge.

The duties were also subdivided; Orange were providing online streaming and Liquid to handle the conference connectivity.

But the milkshakes worked well for all attendees….


What do you think of Tandaa content grants? (15 marks)

Since 2008, the government through the ICT Board has invested more than $ 6 million in local content generation and if you count the money for software and other services, then its $114 million, as part of the World Bank fund.

Recently Tandaa awards were announced and you can read more from Techweez and in 2008, I wrote a story on the first Tandaa grants of $ 4.1 million.

What do PS Ndemo and Kukubo feel about content generation? Here is a response I got a month ago when doing research on local peering and internet cost reduction issues. I just copy pasted from the email.

From Dr. Ndemo:

Personally I am still disappointed with the speed at which we are developing local content.  By now we should be having local vernacular wikis as a strategy to preserve our cultures.  Universities are still sleeping withe several shareable research yet others cannot access such valuable information.  You cannot locate many places in Nairobi yet this is a good content proposition.  We cannot have these many opportunities with huge unemployment at the same time.  I am happy that GOK has taken lead with open data.

From Kukubo

I don’t know what the growth figures are as evidenced through bandwidth uploads, but I can suggest that  there is growing local content development.

From a user generated content perspective, Facebook would probably be a great driver. Institutional content continues to grow with open data, development of educational and health content.


I have not interpreted what it but I guess you can do it better.

Moving on……

If I remember well, part of the reasons we were excited about fiber was that it would lower latencies and better still, it would allow people to generate content and such content would be accessible to people.

Yes, Joe Mucheru of Google argued with significant force of reason that it doesnt matter if content is hosted locally or internationally, because the internet is global. Ali Husein, Kenic board member (or former) termed my post as simplistic, because the hosting business is international and Kenyan government has no business promoting local hosting companies. You can read it all in the comments section in this post and you can read another I have done on the disappointment of local hosting.

Back to the $114 million fund. I know I am simplistic but why invest all that money in content and applications then host it abroad? Yes, there are critical applications and websites that must go on the Amazon or whatever cloud because of SLAs and stuff but is this the case in all the content or majority of it.

So, is Kenya hosting that bad? Why haven’t the grants improved that? Maybe there should be a grant to help this…

For instance, today CCK launched this website, where is it hosted? You have guessed right, not in Kenya, maybe the CCK or the UPDK team could not get Safaricom, AccessKenya, Telkom Kenya or Wananchi group to host it free in their cloud. I am imagining at this rate, all that talk of big cloud content by 2015 is looking very bright.

Well, I just could not let this go….. here is a boring conclusion.  Here is the traceroute of the CCK sponsored site. Check how long it takes and the route it takes…… and we were trying to bring down latencies from 600ms…..

traceroute to (, 64 hops max, 40 byte packets

1 (  1.184 ms  0.854 ms  0.654 ms

2 (  4.720 ms  1.476 ms  1.442 ms

3 (  890.652 ms  1033.720 ms  915.801 ms

4 (  780.050 ms  975.362 ms  686.306 ms

5 (  948.681 ms  864.720 ms  836.259 ms

6 (  889.112 ms  909.568 ms  1014.824 ms

7 (  1051.909 ms  582.662 ms  565.692 ms

8 (  904.593 ms  1070.709 ms  1069.041 ms

9 (  1190.310 ms  1078.289 ms *

10 (  908.698 ms *  1350.263 ms

11 (  1177.123 ms (  880.283 ms (  1013.940 ms

12 (  1174.197 ms *  998.332 ms

13 (  842.497 ms  864.748 ms (  1192.667 ms

14 (  1387.871 ms (  1493.622 ms (  1246.973 ms

15 (  1237.928 ms  1067.814 ms *

16 (  1100.651 ms * (  1349.609 ms

17  * (  1428.182 ms *

18 (  1464.183 ms *  1434.943 ms

Safaricom’s failure to invest in fiber may be slowing its competitiveness

When safaricom announced that they were not making money on some of the data subscriptions, the question on many people’s mind was “how, with all that money and infrastructure?” For others it may have been that Safaricom wants to milk the last of your coins.

Well, apart from the modem and handset data bundles, Safaricom’s other data subscriptions are way high, the cheapest is 11,000 compared to Zuku, 1,500 and Access Kenya, 4,000. The others may fall in between. Most data services require fiber, whether its last mile, interconnecting Base stations or redundancy, fiber is critical.

Ever wondered why Safaricom hasn’t given Zuku a run for its money? Given the financial muscle, you would have expected Safaricom to be competing on the same level withZuku and Access Kenya.

Well, for starters, Zuku’s Wananchi Group owns most of its fiber to the home- but this is mainly in major towns and they have no extensive wimax network compared to Safaricom. On the other hand, Safaricom is dependent on Jamii Telecom, KDN, Telkom Kenya and National Fiber Optic Backbone (NOFBI) to deliver services. For Safaricom to make money, it will have to take into consideration the costs charged by other operators, then put their margin. This means that the costs may be higher than other operators like Telkom but it also depends on calculations.

To be clear, for the 3G network, Safaricom is in pole position because they have the extensive infrastructure courtesy of their base stations but for data, they are limping while consumers crave cheaper or affordable options. That is why when they introduced the unlimited bundle, individuals and small businesses came rushing in.

I was surprised that Safaricom gave an affordable 3G based option that seemed to suit SMEs and expected people not to misuse. Truth be told, I am not sure whether it is a marketing issue or not but Safaricom seemed to attract all the customers including those who misuse, while competition (read Orange) has similar unlimited offer but doesn’t complain of misuse.

As part of my conversation with Rita Okuthe, head of customer proposition at Safaricom, I had to ask the technical infrastructure and why cost to the end user isn’t falling. There is all the talk about fiber this and that and yet we cant seem to pay less. Read part 1 in the earlier post.

Back to the topic. What are the tech issues?

Late entry into laying fiber

Five years ago, companies were busy buying stake in TEAMS and EASSY and others laying terrestrial fiber. Safaricom has a stake in TEAMS but so is JTL. But you got to give it up to JTL because they went ahead and laid metro fiber in all major towns, then made sure that they signed with Safaricom as their anchor tenant.

That was a good idea because JTL did not seem to compete with Safaricom, they would just provide the infrastructure and maintain and Safaricom would pay rent. At some point Safaricom was on KDN but we all know the story with KDN deserves a blog post, not a paragraph.

Anyway, it all seemed to be working ok until Safaricom started getting a lot of clients and realised that the costs were high. After dislodging KDN, guys at JTL also grew horns that I am told a simple fiber survey to Safaricom takes a month and mostly because the guys at JTL have to do it. They move at their pace and there is probably nothing much you can do because options may be limited.

Safaricom may argue that they are still able to make money but late last year, the company invited tenders for laying of its national fiber infrastructure, which will give it more control and leverage to compete with the other players in the industry.

I remember asking Bob sometimes back whether it means Safaricom was getting a raw deal from JTL and whether they will migrate all the capacity. He said the they were just increasing the options because they were also using other providers but if you have your stable infrastructure most of the traffic will go there and others will provide redundancy.

A major challenge of having no fiber is carrying their capacity between Nairobi and Mombasa for onward transmission by global carriers. Two weeks ago during a press conference, I recall Nzioka Waita, Corporate affairs director at Safaricom saying that the high cost of a link between Nairobi and Mombasa was contributing to the high costs of internet access.

Just to put it into perspective, KDN charges $300 per Mb between Nairobi and Mombasa, $420 for Nairobi London, and $420 for Mombasa and London. I couldn’t get any figures from Safaricom because its the closed period but I guess the figures are indicative that the others may not be far off. Just like in any other sector with few players, the prices may not vary on the routes.

The National Fiber Optic Backbone(NOFBI) is different because companies pay $50 for a kilometer of dark fiber then they have to light it. In areas where JTL has proper coverage, I am not sure if Safaricom is opting to light or just go with JTL.

On the last mile, Safaricom may own the infrastructure but in between there are other costs to other infrastructure providers. Compare this with Telkom Kenya which owns its fiber and has terrestrial network and in other areas it is served by the NOFBI, which it operates on behalf of the government.

Because all the base stations are linked with fiber, in some cases microwave, which attracts license fees, you can now see how the cost calculation gets higher for Safaricom.

We discussed 3G technology evolution and how software upgrades allow Safaricom to push more capacity within the same spectrum e.g recently launched HSPA+ but the question of lower costs is a thorny one; if you are pushing more capacity within the same spectrum, shouldn’t that mean lower costs? Well, Safaricom just like other operators will say that they are investing in this technology and the benefits are yet to come.

Growing international content vs. local content exchange

I have been asking this question and I am yet to get someone to give me an idiot centric answer. Make me understand and you can assume or believe I am an idiot, that way, you can start far and drive or walk the point home.

Safaricom says that it doesnt matter whether the content being accessed is local or international, it will still affect the network on the end user or access part and if anything, the costs incurred on setting up and maintaining the network do not equate to any savings that may be saved by local peering (exchange of content between ISPs).

My argument is that you pay $300 per Mb on the international transit, if 1,000 of us can save the company that by hosting locally (assuming it works), then you can reduce maintenance costs and pass them down to end users. I keep giving examples of South Africa where locally hosted content is relatively cheaper because its lesser cost on the ISP.

Even though Safaricom owned up that most of the content accessed on the unlimited offer was international, it was argued that there was no way for a consumer to know whether the site is hosted locally or not, and therefore the cost may not be an issue for the customer.

My argument was that I may be an idiot but I will notice that on month one the bill was lower and the other was higher and if the reason is that I was playing games hosted locally vs hosted abroad, then I will have a choice. Again, I may be an idiot but I surely know the sites I visit.

Yes, infrastructure has grown locally and some of the content like Youtube videos may be available locally through its global cache shared at KIXP, but is this it? Does it mean local content has no role to play in lowering connectivity costs in the country?

Safaricom has invested more that $ 2.4 million (sh. 2 billion) in their cloud service. Let us see how the marketing exercise unfolds.

So, it may not be a proper explanation for Safaricom data costs, but looking at it from their side, you may understand why. But I may be wrong.

Orange, Safaricom offer new 3G speeds

This week, Orange and Safaricom were busy launching higher speeds on the 3G network. As usual, Safaricom was the first to launch and few days later Orange did.

Safaricom invite said that the speeds were increasing from 14 megabits per second (mbps) to 21mbps on their 3G network. I remember a friend asking me whether I have ever hit 14mbps on 3G and to be honest I have never, the best I have got is 7mbps and given that I had used the modem from the days of 2G (Edge) I was happy.

The problem is that no one tells us that when we say 7mbps, thats the speed that you will probably not ever reach, the average speeds will be 3mbps or something, that is why they do tests. So, dont expect to hit 21mbps yet on your modem.

For the sake of clarity, this post will address 3G on GSM and CDMA, which is currently being offered by Orange. The two technologies use 3G, only that one has better experience for voice and data while the other one is not exactly nice.

CDMA (code division multiple access) is mainly known as the landlines that we carry around, Orange provides a 020 number which comes in handy for landlines calls, its probably the reason Telkom Kenya stopped extending its copper and ADSL service, which can be addressed in another post.

CDMA uses 3G and Orange provides speeds of 3mbps download and upload of 1mbps, but if you are more into voice and data on your handset, it does not give a good experience especially if you are used to GSM. Use it for data or voice separately and you will enjoy otherwise you will not have good things to say about Orange. I have tried it and had to trash it.

So, what does it mean to subscribers?

In technical terms, it means that Safaricom and Orange have upgraded their GSM networks from High Speed Packet Access (HSPA) that supports speeds of up to 14mbps to a higher version, maybe HSPA+ which can support speeds of up to 42mbps (no confirmation of platform yet).  Theoretically, 3G networks reach max speeds of 7.2Mbps down and 1.4Mbps up but later technologies improve on the speeds e.g HSPA.

Out of 1500 3G sites, Safaricom says  80 sites are upgraded and users can enjoy faster speeds. Orange says it has 220 3G sites but its not clear how many have been upgraded.

This means that now the modem will be accessing internet faster and pages will be opening faster.  This will cost more or less depending on how you use the net. For those in cyber cafes or paying per shilling on the mobile, you may use lesser money because the pages load faster.

For those paying per megabyte (MB) may not see signficate change but this will depend of the websites accessed e.g if you access videos, they are heavier compared to plain sites like

The cost will also depend on whether the sites are hosted locally or abroad, local sites may be make the cost lower because they load faster and other video sites like You tube may load faster because Google has hosted some of that content locally. Read more about it here.

There has been constant complaints that the modem is expensive; this maybe the case but people are also careless, we leave websites downloading even when you have accessed the content you were looking for. For instance, at the lower end of your browser, you see a lot of stuff going on like 5 out of 15 downloading, if its at six and you have gotten what you were looking for, click on the stop button and you will save on the MBs.

Orange will argue that at the cost of 4o cents per MB, you do not need to stop the download because its cheaper but then, it depends on our spending habits.

The other thing is a choice of whether the modem works for you or you need another internet service. If you use Ksh. 3k or more on mobile data or on the modem, you need to explore other options that will allow better experience.

So, enjoyed the speeds? Share with us your experience.