Day 1 – Sunday, 22 September 2013
Welcome
Philip Smith, MENOG Coordination Team
RIPE NCC Regional Meeting Welcome
Marco Hogewoning, RIPE NCC
Keynote
Qusai Al-Shatti, Vice General Manager, Central Agency for Information Technology, Government of Kuwait
This presentation is not available online.
In this keynote presentation, Qusai Al-Shatti highlighted the need to transition to IPv6 in a post-IPv4 depletion world and outlined steps that network operators and others can take to start making the transition. He highlighted the need to establish regional IXPs so that ISPs across the Arab region can peer directly with each other and attract more global content aggregators and providers. He also reviewed some of the different IXP models, both commercial and non-commercial, that have been used by existing IXPs.
Jan Zorz, Internet Society (ISOC), asked about the status of the Arab region in terms of IPv6 implementation, and commented that NAT is just delaying the inevitable.
Qusai answered that although there is interest in IPv6, it’s small. He said the answer is to encourage collaboration between all stakeholders. He mentioned that the EU has a goal of 25% IPv6 deployment by a certain date and that the Arab region needs to set a concrete target of their own, perhaps 5-10% by a certain date. Currently, he said, the Arab region is still using mostly IPv4 and that IPv6 take-up is slow.
Mohamed El Bashir, ictQATAR, In Qatar, publishing national IPv6 strategy and plan agreed with stakeholders and taskforce. In UAE, also have a taskforce, so there’s a push from government operators, but network operators are key to pushing implementation. Pushing operators from demand side is a workable solution.
Ali AlShamali, Zain Kuwait, commented that all operators are considering an IPv6 strategy and that NAT is no longer efficient. He added that Zain Kuwait is moving towards IPv6 and that they have a strategy in place for next year.
Qusai asked what the timeline of their plan was.
Ali answered that their plan has been laid out for next year. He mentioned that there’s an issue of security once you remove NAT, but that Zain Kuwait is actively engaging in the matter.
Qusai answered that he was happy to hear that.
There were no further questions.
Technical Keynote: Internet Development Lessons and Experiences
Philip Smith, APIA
This presentation is available online at
http://www.menog.org/presentations/menog-13/198-internet-development.pdf
Philip gave a history of the early days of the Internet and the technical issues involved in switching from classful to classless BGP routing. He reviewed the issues that classful routing had left behind and the improvements that have taken place in BGP since that time. He talked about the early days of peering and establishing one of the first IXPs at LINX, and reviewed the lessons learned from LINX and others that followed around the world.
Joseph Jabakhanji, TCL, asked how regulation worked in countries that had no rules or laws to govern the Internet.
Philip replied that is depended on who the regulator was. Many looked at the telecommunications market and tried to apply that to the Internet market. He mentioned a positive example in which the regulator was very open to market competition and so it was an easy conversation to have. But he said it depends on the circumstances and that you can’t just tell them how it will be. Instead, you have to understand the local issues and circumstances and it has to be an ongoing dialogue in order to understand what they’re trying to achieve.
There were no further questions.
Lessons Learnt Deploying Global Dual Stack and DDoS Networks
Raju Raghavan, Tata Communications
This presentation is available online:
Lessons Learned and best practices – Engineering a global dual stack and DDoS Mitigation infrastructure
Raju shared his experiences at one of the world’s largest ISPs and focused on the technical side of planning for traffic growth and customers’ zero tolerance for downtime. He also shared best practices for dealing with DDoS attacks.
Massimiliano Stucchi, RIPE NCC, asked whether there was any reason why Tata Communications uses IS-IS as opposed to other protocols.
Raju answered that they have global networks and use different protocols in different parts of it. He said that IS-IS has definite advantages with IPv6 because of its multi-topology.
Philip asked about using IS-IS versus OSPF.
Raju answered that IS-IS is easier to deploy and that there are layered announcements happening over IS-IS, so he thinks it’s a good choice. However, he added that it depends on your network.
There were no further questions.
Security Scenarios in IPv6
Massimiliano Stucchi, RIPE NCC
This presentation is available online at
http://www.menog.org/presentations/menog-13/197-IPv6-Firewalling.key
Massimiliano explained how, as a RIPE NCC Trainer who introduces a lot of people in the RIPE NCC service region to IPv6, he gets a lot of questions about IPv6 related to security. He gave an overview of security issues and best practices concerning IPv6, which he emphasized will only become more important as we move towards realising the Internet of Things.
Jan Zorz, Interent Society (ISOC), commented (as an IPv6 user and not in his ISOC role) that the presentation was brilliant, but that the ‘customer assignments’ slide about the number of subnets needed for home users and business users was harmful. He said that some end users could require a larger allocation than a /64, and if users start searching for ways around this limitation, it could lead to major problems. He added that although a /56 for home users and a /48 for business users is the standard recommendation, his personal opinion is that everyone should receive a /48.
Massimiliano answered that the delegation sizes in his presentation are not what the RIPE NCC recommends, but what he sees happening in the real world. He said that the RIPE NCC trainers recommends check… He stated that he agrees that a /64 is not enough for some users.
Philippe Nahas, datamena, asked what kind of growth Massimiliano sees in IPv6 in the RIPE NCC service region. He also asked when a typical LIR should start worrying more about security than adoption.
Massimiliano said that, speaking personally, he thinks IPv6 adoption is going very slowly. RIPE NCC trainers ask attendees at training courses, and find about 10% of attendees have any experience with IPv6. This is why the RIPE NCC encourages operators to attend the courses. He said that security should be a main concern right after implementing your IPv6 network, and that’s why Cisco did it right away. He understands that the problem is having to teach your support to handle IPv6 issues, but that the best case is to implement security as soon as possible.
Philip Smith, APIA, commented that his experience in the Asia Pacific region is that network operators in that region are implementing IPv6 security right from the very beginning. IPv4 security measures are being replicated in IPv6 and will block IPv6 access if those security features are not yet in place. These network operators are being very cautious, with their IPv6 rollout planned in step with security measures. He said that it’s best to replicate what you have in IPv4 at the same time as you rollout IPv6.
There were no further questions.
Detecting and Preventing IP Route Hijacks
Marco Hogewoning, RIPE NCC
This presentation is available online at
http://www.menog.org/presentations/menog-13/194-MENOG13-Hijack.pdf
Marco highlighted the importance of the RIPE Registry and reviewed the function of a registry once IPv4 has run out and how it can be used to improve network security using RPKI. He also emphasised the importance of keeping the RIPE Registry up to date.
There were no questions.
Routing Table and Deaggregation Report
Philip Smith, APIA
This presentation is available online at
http://www.menog.org/presentations/menog-13/190-routing&deaggregation.pdf
Philip gave an overview of the history and trends in allocations and announcements, along with interesting trends that have appeared over time. He also delved into the CIDR Report and Routing Report on deaggregation and explained the deaggregation factor for different regions, as well as the BGP Instability Report. He emphasised the importance of aggregation and demonstrated how lower standards in different regions can harm the entire Internet.
An attendee asked how to convince network operators to aggregate when they’re breaking down allocations into smaller prefixes.
Philip answered that newer ISPs use the “shot-gun” approach of chopping up /10s into /24s, and added that if everyone did that, the Internet wouldn’t work. He said many operators have started filtering out these /24s, although how to do that without breaking valid routing is difficult. He said that very few providers really do need to chop up their allocations into /24s, and there is almost always a solution that gets around having to do this. It’s best to sit down with consultants and figure out a solution that works for you. It takes a lot of work and there is no magic solution, but there are a number of people willing to help and try to make a difference, such as himself.
Jan Zorz, Internet Society (ISOC), asked about the IPv6 CIDR Report, and Philip answered that he’s been working on it for five years.
Jan asked about aggregated savings on the same AS path, and Philip answered that that is the best possible aggregation.
Jan asked if Philip ignored traffic engineering and Philip answered that he did.
Jan suggested that some operators deaggregate because they don’t know how to do things better, some do it to facilitate traffic engineering, and some do it out of fear of hijacking. He asked Philip how many fall into each of these three categories.
Philip replied that, in his experience, most network operators claim they deaggregate to facilitate traffic engineering, but that when Philip looks into it, there hasn’t been any true engineering taking place. They would look at all /24s and send them where needed to cover broken links (so random directions to balance traffic). The people doing real engineering seem to know what they’re doing – sub-aggregating, etc. Only met two providers where engineer said they were on board but management specified /24s to counter hijacking.
Jan suggested that a tool to help aggregate is needed.
Philip replied that there was software written based on AS path (identical AS path and would drop prefixes). Fear about this among ISPs, though, so tough to get everyone onboard.
Fredy Künzler, Init7 (Switzerland) Ltd, said that he had been preaching against deaggregation for a while now. He wanted to share what he just learned, which is that some regions, like the Arab region, don’t have routing capabilities in Europe where they pick up their links, so they deaggregate to avoid packet loss. With these links in place, traffic engineering would be much easier.
Marco referred to the keynote address from the morning’s session, when Qusai Al-Shatti pointed out the need to establish regional IXPs so that routing wouldn’t have to go through Europe.
There were no further questions.
Story of RIPE 554
Jan Zorz, Go6 Institute
This presentation is available online at
http://www.menog.org/presentations/menog-13/186-Jan_Zorz-ripe-554.ppt
Speaking in his role as CEO of the Slovenian Go6 Institute, Jan gave an overview of ripe-501 (later ripe-554), which began when he questioned the Slovenian government about requirements for IPv6-capable equipment. Several years, many drafts and much community involvement later, the RIPE Document, “Requirements for IPv6 in ICT Equipment”, has now become the standard for IPv6 equipment procurement.
There were no questions.
Unknown Arguments for Peering
Marco Brandstaetter, DE-CIX Management GmbH
This presentation is available online at
http://www.menog.org/presentations/menog-13/204-Presentation_Unknown_Arguments_for_peering_MENOG_2013_09.ppt
Marco reviewed some of the well-known arguments for peering, including that peering is cheaper than IP transit and lower latency. But some of the unknown arguments include recognition, collaboration, more revenue through peering, and finally, that the UAE-IX is available.
Alaa Abu-Habib, FastTelco, asked whether UAE-IX had any plans to come to Kuwait for the Kuwait Internet Exchange?
Marco replied that no, the concept is that every data centre will be connected through UAE-IX, which they want to be the regional IXP. However, he said, UAE-IX supports and is happy about every local Internet Exchange Point initiative, because they believe they bring the whole region forward.
Marco added that no matter how you’re connecting, where you’re going through, etc. , you can peer via UAE-IX. He added that this is not a model that’s been seen in the region so far. He said he thinks the future will include freer models, in which everyone is more free to connect – like in Frankfurt – and that he thinks this has real added value.
Philippe Nahas, datamena, sai he sees many peering openly and asked whether that is the same trend as in Frankfurt and at DE-CIX.
Marco responded that they have open and selective peers and that at UAE-IX, they did traceroutes and saw that it would also be good for tier 1’s to peer there, and then they would defintely be restrictive peers.
Philippe asked Marco to show his traffic graph again, and said he had heard from his colleagues that the big spike shown in the graph represented substantial financial savings and so there was obviously real value in peering.
Samir Sogay, Gulfsat Communications, asked whether the UAE doesn’t already have a similar Internet Exchange Point, and how UAE-IX plans to compete with it.
Marco responded that he didn’t believe there was another Internet Exchange Point using the same model as UAE-IX, including being carrier neutral, data centre neutral, and connecting everyone at no cost. He added that although he likes what exists there, it is not the same as UAE-IX.
There were no further questions.
Network Traffic Telemetry
Paolo Lucente, pmacct
This presentation is available online at
http://www.menog.org/presentations/menog-13/195-Lucente_network_traffic_telemetry.pdf
Paolo gave an overview of the NetFlow, sFlow and IPFIX network telemetry protocols along with recommendations on how to venture into this area. He explained how telemetry protocols are being generalised and that additional pressure is being put on the collector. Tips included employing a top-down approach and having specific, well-defined goals in place in order to keep things manageable.
Marco asked who uses sFlow or NetFlow and several audience members raised their hands.
There were no further questions.
Best Current Operational Practices Efforts from the Internet Society
Jan Zorz, Internet Society (ISOC)
This presentation is available online at
http://www.menog.org/presentations/menog-13/185-Jan_Zorz_ISOC-BCOP-v.4.3-MENOG.ppt
Jan gave an overview of ISOC’s Deploy360 program, which offers resources for network operators and other stakeholders in deploying and operating IPv6 networks. He highlighted the need for a credible repository of IPv6 best practices documents for network operators around the world.
There were no questions.
Traffic Engineering with Excel: Commercial Aspects of IP Capacity
Fredy Künzler, INIT7
This presentation is available online at
http://www.menog.org/presentations/menog-13/200-21092013_MENOG13_Traffic-Engineering_with_Excel_Fredy_Kuenzler_Init7.pdf
Fredy identified some myths that prevail when buyers make decisions about IP capacity, including that transit and public peering are the only options, that tier 2 is inferior to tier 1, and that public peering reducing transit costs. He then outlined a better strategy for ensuring sufficient capacity for end users that will make your CFO happy.
Jan Zorz, Internet Society (ISOC), asked for how many years you can keep on reducing costs.
Fredy replied that submarine transport is the biggest cost and that will still rise, but he thinks it will eventually be comparable to transatlantic costs. He said that capacity is expensive in the Middle East for various reasons, and that has to be corrected some day by the community itself in order to be sustainable.
Jan asked (jokingly) whether the goal then was to cut down on costs a little every year, but not too much, so that costs could be reduced indefinitely.
Fredy replied that the goal is good quality for the end user. He said that ‘business types’ care about the financial calculations and they’re the ones who have to deal with that.
Marco Brandstaetter, UAE-IX, said that the future outlook of the costs of STM-64 assumes Europe is the centre of the world when it comes to the Internet, and that perhaps he’s too enthusiastic about peering and what’s happening in the Arab region, but he doesn’t think so much traffic through Europe is necessary. He also pointed out that there are Arabic parallels to many content providers, such as an Arabic Netflix and Arabic Hulu, etc. He said he thinks more traffic will originate regionally in the future and that will help reduce costs.
Fredy said he agrees with that, but said that transit costs are still too high in the Middle East (10 or 50 times higher than in Europe). He suggested that operators should sell their excess outbound bandwidth for a cheap price to help facilitate more regional content.
Marco suggested the problem could also be solved by engineering. He said the main reason people host content in Europe is the pricing, and that peer-to-peer solutions offer a cheaper alternative. Until now, he said, operators have been restricted to a handful of nearby ISPs, but now there are other options emerging that offer broader opportunities.
An attendee from KuwaitNET said that they used to host servers in the US and then the Netherlands, and that it wasn’t about saving money but latency. When they were asked to move to the Middle East region, he said, the prices were very expensive. He added that it doesn’t make sense to pay triple just to save a few milliseconds in latency, so bringing content inside the region will help a lot. He said that it’s happening, and that having an IXP is a good step forward and will make CFOs happy.
Philippe Nahas, datamena, said he agrees with the approach, but thinks Asian traffic is missing from the picture. He said that growth in that region is staggering, so he would definitely suggest planning for reaching Asian sites in a cost-effective way in the future. He also asked Fredy to clarify a point from his presentation regarding how tier 2’s can offer a cheaper alternative than a tier 1 supplier.
Fredy responded that, if you look at how traffic originates, the rest of the world you have to reach after using a tier 2 is small. He suggested someone could buy the small chunk they need from a tier 1 to fill in the gaps left by exotic locations they can’t reach in their routing table. He said that you can cover a lot through peering for free or at low cost, and then buy the small remaining pieces from a supplier.
Philippe pointed to Fredy’s claim that peering is cheaper than transit at sub-10-gig levels, and asked how that was calculated.
Fredy responded that, when you add up all the other costs, you realise that the level at which peering starts to pay off continues to increase. He said that if you just want 4-5 Gb traffic, you’d be better to sign up with a network that already has peering set up.
There were no further questions.
Day 2 – Monday, 23 September 2013
IDN (Internationalized Domain Name) Panel
Panel Chair:
- Bashar Al-Abdulhadi, KuwaitNET
Panelists:
- Baher Esmat, VP, Stakeholder Engagement – Middle East, ICANN
- Mohamed El Bashir, Technical Manager, Qatar Domains .qa
- Abdullah AlBarrak, موقع. Arabic gTLD IDN operator
Bashar opened the session and introduced the panelists.
Baher Esmat, VP, Stakeholder Engagement – Middle East, ICANN
Baher gave an introduction to IDNs. Like IPv6, he said, people started thinking about the introduction of domain names in scripts (such as those used in the Arab region, India and China) 15 years ago. He said there was a lot of research in late 1990s, then around 2002-2003, ICANN started thinking about how to make DNS understand domain names in different scripts. He said they decided that they don’t need to change DNS, but the protocol that would make applications understand the scripts. Around 2004, he said, IDNs were implemented at the secondary level. He stated that the solution wasn’t perfect, because half the names were in Arabic and half were in ASCII. He said that people wanted full URLs in other languages, and the second phase (from 2005-2009) was to develop the policy around this and implement it. In 2009, he said, they started a fast-track process, which allowed countries to get country codes in native scripts. He added that there are currently more than 30 country code names inserted into the root (consisting of about 23 countries, because some have several languages for one country, such as India). He said there are 11 Arab countries with Arabic IDNs (but not Kuwait). He stated that the full list is available on Ithe CANN website. He added that there hasn’t been much uptake on IDN registration of domain names in the Arabic region due to the lack of content in Arabic.
As for generic top-level domains (gTLDs), Baher said, there is a new program that was launched by ICANN in 2011 to add more TLDs. He said there were approximately 1,900 applications received since 2011, including 116 IDN applications. It is difficult to foresee how those names will do, he said; some will probably do well and others less so, because it all depends on the business plans and policies involved.
Baher said there are two overarching issues that need work in the area of IDNs: variance and universal acceptance. Variance, he explained, refers to the problem of having names that are similar to one another. He stated that this is common in some scripts, including the Arabic script (because there are several different languages that use the Arabic script). Characters in some scripts, he said, look the same but are considered different by the computers reading them. He said that ICANN is working on reducing variance in the root and that they recently set up an expert panel to look into the technical issues and perhaps develop policy rules to introduce variant characters into the root, working on 15 different scripts. He said they are reaching out to experts from different communities (technical, linguistic, etc.) to take part in that work and invited the audience to get in touch if they were interested in getting involved.
Baher explained that the second problem, universal acceptance, is not new. He said that when new gTLDs were introduced a decade ago, those names were new to the root and some had more than three characters. They then started to see some online applications not accepting these, he said (for example, when entering email addresses with a TLD containing more than three characters). He said they will face this issue again with thousands of new TLDs being introduced in the next few months and that ICANN, in cooperation with the community, is working on ways to resolve this issue. He added that recently, there was a report recommending ICANN update its IDN guidelines to encourage DNS operators and other service providers to adopt new guidelines, which encourage using the root database as the only authoritative reference to verify TLDs (and not use hard-coded lists, for example).
Baher reiterated that variance and universal acceptance are the two main issues they’re dealing with. He then pointed out an example of the variance problem (using the AlSaudiah domain name) from a list on the ICANN website.
Mohamed El Bashir, Technical Manager, Qatar Domains .qa
Mohamed explained the issue from a registry point of view. He pointed out that Mac is not Arabic-friendly (scripts appear scrambled), and said that this is just one issue. He said the demography of Internet users is changing, and that there are 2.2 billion end users online, with almost 591 million Chinese (22% of total) users. He added that Arabic growth is huge, especially after the Arab Spring. He also pointed out that there has been big growth is the use of social media. For the next generation of Internet users, he said, English won’t be their first language, and that we need to tackle this challenge on a technical level. He said things are now changing at the gTLDs level, but have already been changing at the country-code level for some time (for example, for the past three years this has included Cyrillic, Arabic and Chinese scripts).
Mohamed suggested this is the biggest Internet change since its inception. He said that 1,900 new gTLD applications have been received by ICANN, and that there will be domain names for “dot anything” (such as .music, .help, .cam, etc.). He said that no one knows exactly how this will affect the way people access information online. He summarized the timeline of events by stating that gTLDs were first, then ccTLDs, IDN ccTLDs, and IDN gTLDs.
Mohamed said that, in terms of global IDN registrations, the Russian registry has been the most aggressive, with 800,000 registrations, then the Chinese, with 275,000. He said about 30% of Russian registered names are active (so there is defensive registration happening).
Regarding Arabic domain names, he said, there are more than 10 delegated Arabic IDN ccTLD strings in the root (Egypt, Saudi Arabia, Emirates, Qatar, Jordan, Tunisia, Morocco, Algeria, Syria and Palestine). He said there are linguistic and technical challenges involved in the Arabic language. The linguistic challenges, he continued, include the fact that the Arabic language is complex and rich – there’s a huge variation in words, the Arabic script is used by several different languages (Farsi, Urdu, Arabic, etc.), there are 28 characters that can take on different shapes depending on their location in a word, and there are two different groups of numerals used. Of the technical issues, he said, the main one is usability (i.e. global acceptance). He said that browsers don’t accept the Arabic script (they often convert it) and that search results are problematic, but that they’re starting to see improvements from Google. Arabic emails, he added, is another problem, and social media networks generally don’t support different scripts. So, he concluded, we need to have applications to support IDNs, not just registries.
Qatar Domains Registry has three domains, he said: .qa, .doha and one in the Arabic script. He added that Qatar Domains Registry uses a competitive model and that they have about 15 registrars now (three local and 12 international), so.qa can now be registered from anywhere. He added that they have 17,000 registrations at this point.
He said the technical challenges related to IDN experienced by the registry, if anyone wanted to set up their own, included needing to have the support of the registry system for Arabic IDNs, needing language tables, coming up with solutions to the variance issues (i.e. deciding whether to bundle similar names or not, which is a sensitive issue).
He said they launched the Pioneers Campaign, a publicity campaign that tried to convince big websites to use the Arabic script in their domain names (for example, the University of Qatar, newspapers, etc.). Qatar’s particular challenges, he said, are that the population is 1.8 million, so it’s a small market; it has highly educated Internet users who are already comfortable using English and can be difficult to convince to switch to Arabic; and that more application and software support is needed (which is a universal challenge).
Mohamed stressed the importance of marketing and awareness building (they had magazine ads, booths at events, etc.) and said that their campaign was based on the concepts of trust, availability and identity.
He added that they recently launched a mobile app (“Qatar Domains”) that allows users to immediately see whether their domain name is available, along with ownership details of registered domain names.
So, he concluded, their solutions revolve around joint research projects (academia and technical collaboration); resolving issues of usability and support (software and applications); marketing and awareness; and innovation and entrepreneurship (content development).
Abdullah AlBarrak موقع. Arabic gTLD IDN operator
Abdullah, via Skype, gave a brief introduction, stating that they have been part of Arabic domain names since 2000 and developed lots of content with ICANN. He explained that they tried to get a gTLD that could be used in other languages using Arabic script, and that they will hopefully register it next month.
Basher then opened the panel discussion with the three panelists and invited comments and questions from the attendees.
Reyadh A. Alkazemi, Ministry of Communications, asked (speaking personally), whether we need everything in Arabic. These days, he said, it makes it difficult to search, etc. He added he has switched from Windows to Apple and is facing many difficulties. He asked: If we use one language for everything online, wouldn’t it be easier? Why do we need different languages? He said that, although we want to support our language, let’s go with the trend of the rest of the world. Maybe China or India need to use their own language online because there are so many end users, he said, but it’s not the same in the Arab region.
Baher responded that the question of complexity is not new; whether we need this has always been part of the debate. But the reasons for doing it in India or China apply in Arab region, too, he said, and that even if people can speak English, they may be more comfortable using Arabic. He added that statistics show many mobile users and chat rooms use Arabic, so there is a need to make the Internet accessible to people in their native language. Whether it’s required at the domain level is another discussion, he said, and can be debated. He added that he must admit that part of the motivation for IDNs is political; governments see value in having their country names in Arabic. But, he concluded, there was still demand from end users.
Mohamed responded that there has been a huge explosion in social media use in Arabic, although he admitted there had been less used in academic and Wikipedia content. Still, he said, there is huge demand from end users, and we also need to make the Internet accessible to all types of users, including those who are less educated and don’t know as much English. The magnitude may be different in China, for example, but the need still exists in the Arab world, he said.
Abdullah responded that the Internet is for everyone, not just the well-educated who speak good English, as is common in the Arab region. He also said that there are so many interpretations about how to spell Arabic words in English, so this can actually be more confusing than simply using the Arabic.
Bashar concurred that the use of Arabic online is rising.
Bashar asked Abdullah why they selected the domain name they did.
Abdullah responded that they wanted to use the most generic and obvious domain name they could. They wanted to find the .com of the Arab language, he said.
Bashar asked Baher why they selected .arab as the domain name.
Baher responded that some governments wanted to have a presence in the Internet space and have .arab as one of the future TLDs. He explained that they decided eventually to apply for two strings: one in ASCII (English) and one in Arabic. He said they thought the typical translation was .arab (as opposed to .arabi, for example). He added that he doesn’t know how they plan to use the name, but will involve other Arabic organisations in the decision-making process.
Mohamed responded that domain names are like real estate – they need to be unique in order to have value. He said they have translations of TLDs in Arabic, and that which is most attractive to users depends on how they’re used. He said he thinks there will be lots of trademark protection, and that branding will have a role, too. For example, he said, .me (Montenegro) is being branded as “dot me”. He asked how we will remember hundreds of thousands of domain names and suggested that the registries should be creative and innovative about how to use the names. He concluded that no one knows how this business venture will end up.
Reyadh A. Alkazemi, Ministry of Communications, thanked the panelists for their answers. He then asked about the cost involved in the marketing required to promote these domain names. He also asked how to distinguish between actual users in the Arab region (for example, people may have multiple mobile devices). He then asked how much we are spending on IDNs and how much income it generates, and whether this is it cost effective.
Mirza Junaid, FASTtelco, asked why the panelists were not speaking Arabic, and asked whether we are separating the Arab world from the West. He mentioned that a session from the previous day encouraged peering and merging.
Mohamed responded that he has given this presentation in Arabic before, and that it’s not about separation, but about going with users’ and companies’ preferences. He suggested you can have both languages, that it’s just about using the right language in the right region or for the right purpose.
Baher responded that, regarding the separation issue, it depends on the audience you’re targeting. In the Arab region, he said, they have media channels and websites in several different languages and that it’s the same thing with domain names.
Marco Brandstaetter, UAE-IX, said that he’s from Austria, where many websites are just in German, while in the Czech Republic there are sites just in Czech, so if content is constrained to just the Arab region, there’s no problem in just using Arabic. If it’s international content, though, then maybe English should be used, he said.
Bashar asked the panelists about moving forward with a ccTLD for Kuwait.
Mohamed responded that when ICANN opened the door for new applications, they were in a good position because they were running the ASCII registry for .qa. But, he added, there’s no requirement that the ASCII and other language registry have to be the same. He said it’s not a huge investment; some countries use open source systems or run their registries manually, so there isn’t a huge cost barrier. He said you just need to fulfill the basic DNS requirement and apply to ICANN. In his experience, he said, it took them almost a year for their own, but now it should take less time because IANA and ICANN are more familiar with the process. But policy development is big, he added, and this will take some time. He said that they are ready to help move a Kuwait domain name forward based on their experience.
Baher said that Mohamed had covered just about everything. He added that the management of the IDN ccTLD registry is up to the region to decide. He said the challenge is not the technical challenge of running the registry – it’s about spreading the word and getting the IDN ccTLD used. There’s a lack of understanding in the industry in this region, he suggested, and ICANN needs to strengthen understanding here to help move things forward.
Bashar concluded the session and jokingly remarked that Microsoft doesn’t support scripts on Mac, so blame Microsoft rather than Mac. He thanked the panelists.
Middle East Engagement Strategy
Baher Esmat, ICANN
This presentation is available online at
http://www.menog.org/presentations/menog-13/199-MENOG13-MEStrategy-Sep13.pdf
Baher gave an overview of ICANN’s activities and plans in the Middle East in order to more fully engage with this region. These activities focus on DNS security and stability, the domain name industry, and the Internet governance ecosystem.
There were no questions.
RIPEstat & RIPE Atlas
Christian Teuschel, RIPE NCC
This presentation is available at http://www.menog.org/presentations/menog-13/188-MENOG13_CT.pdf
Christian gave an overview of two of the RIPE NCC’s tools: RIPEstat and RIPE Atlas. RIPEstat is an all-in-one toolbox that retrieves information about a particular Internet number resource (along with related information for domain names and countries). RIPE Atlas is a global active measurement network that employs thousands of probes measuring Internet connectivity and reachability to provide an unprecedented understanding of the state of the Internet in real time. Christian also gave some real-world examples of how each of the tools can benefit users, such as informing peering decisions.
Jan Zorz, Internet Society (ISOC), asked about the observed BGP peers data and suggested it would be really useful to split those results into observed IPv4 neighbours and observed IPv6 neighbours, in order to facilitate easy comparison and see how much IPv6 has been deployed by an AS.
Christian responded that he thought that was a good idea and thanked Jan for his feedback.
Marco Hogewoning, RIPE NCC, asked for clarification about the data shown in the ASN Neighbours widget.
Christian replied that if you query for a specific AS, the widget will look up all AS route paths and then determine if it’s a “right” neighbour (generally downstream) or “left” neighbour (generally upstream). The developers wanted to stay neutral with regards to what constitutes upstream and downstream providers, he said, and so chose to use the “left” and “right” terminology instead.
An attendee asked what the power value in the ASN Neighbours widget denotes.
Christian responded that different route collectors are distributed around the world collecting real-world data and this is then compared versus the routing tables. The power value shows how many collectors have seen the ASN.
An attendee asked what the requirements are to host a RIPE Atlas probe.
Christian responded that there are no real requirements other than keeping the probe online as much as possible.
Philip Smith, APIA, mentioned that his home probe has been connected for years and doesn’t require any maintenance.
Fredy Künzler, INIT7, mentioned that the algorithm for the left and right neighbours appears to be wrong when testing his own AS.
Christian asked Fredy for more information after the session.
There were no further questions.
RIPE NCC Registration Services Update
Marco Schmidt, RIPE NCC
This presentation is available at http://www.menog.org/presentations/menog-13/202-MENOG_13_update.pdf
Marco gave an overview of the RIPE NCC’s policies and procedures regarding obtaining IPv4 and IPv6 address space, regional statistics, transfers of address space, and how to contribute to the development of address space policy.
Fredy Künzler, INIT7, asked Marco about the IPv4 available pool and whether he knows how long there will continue to be IPv4 space available.
Marco said that at the (recent) rate of 2,000 allocations per year, there will be space available for another seven years. However, he added, it’s difficult to forecast the membership growth (i.e. new LIRs).
Fredy asked how long we want to have IPv4 address space left and said he would rather have it run out completely within two to three years in order to force people to use IPv6. He said he thinks this should be discussed within the Address Space Working Group discussion list.
Marco responded that policy has to be followed, and that there will continue to be new companies entering the market that require IPv4 for some time, so IPv4 will have to continue to co-exist with IPv6 for now.
Jan Zorz, Internet Society (ISOC), asked what the millions of reserved IPv4 addresses are for.
Marco responded that these addresses represent temporary assignments as well as returned address space, which remains in quarantine for some time before being returned to the available pool.
Marco Hogewoning, RIPE NCC, remarked that any new IXPs that may be established will be able to obtain IPv4 address space.
There were no further questions.
Closing Remarks
Axel Pawlik, RIPE NCC, thanked everyone for attending the meeting.
He also thanked the meeting hosts at KuwaitNET, Bashar Al-Abdulhadi and Yaser Mohamed. He also thanked the sponsors and the MENOG coordination team, including Sandra Gijzen and Philip Smith. Axel mentioned the upcoming RIPE Meeting in Athens, RIPE 67, and encouraged all attendees to help represent the Arab region by attending the meeting or taking part remotely. He invited everyone to the next MENOG Meeting in Dubai.
Bashar Al-Abdulhadi, KuwaitNET, thanked everyone for attending and expressed his hope that they will be hosting everyone again in the future.
Philip Smith, APIA, reminded everyone of the MENOG mailing list, encouraged feedback, and closed the meeting.