How Does DNS Help The Internet Scale?
Abstract:
The Domain Name System (DNS) is the primary way that we humans interact with the Internet. It translates human readable hostnames into numeric addresses that packets need to be routed around a network. Unfortunately, every time you visit a site, your computer performs a DNS lookup.
For example, when you try to view google.com , your computer will first ask for the address of google.com by performing a DNS lookup before it can communicate with Google’s servers using this numerical address . This slows down web browsers and other applications which perform frequent domain lookups as they wait for the reply from the DNS server which can easily take 100ms or more .
DNS does not only slow down your web browser . Many recent reports have shown that DNS is responsible for a very significant percentage of Internet traffic, which suggests that optimizing this protocol will result in substantial performance gains. Rather than changing the protocols used to interact with DNS , we propose instead to change how end-hosts use the data received from DNS servers. We present an algorithm called Adaptive DNS that takes advantage of the structure of cached DNS responses to reduce end-host bandwidth consumption by up to 80% . Our system complements existing proposals which seek to save end-host memory consumption at the expense of more frequent lookups .
How does it helps to Scale internet:
Every time you perform a DNS lookup, the client has to send out a request of how it connects to the server. The server has to reply with an actual address or it can say that it doesn’t know which means that they are trying to find that information themselves. This takes up bandwidth when the servers are replying because they have to find out the information that is needed. As mentioned in the abstract, DNS can take up a lot of bandwidth because it has to send out requests and receive answers for those questions.
Every bit counts when you are trying to store large amounts of data or use them online. This saves lots of bytes which make it very useful especially now because people are trying to get more data into less space.
As per the report by APNIC , DNS lookups account for approximately 65% of the request bytes transferred over IPv4 in peak hours. Allowing users to perform fewer DNS lookups will reduce Internet latency, saving time and money.
Enhancing internet’s performance with adaptive dns:
Rather than limiting themselves, as most prior work has done, to the lookups of the frequently accessed domains, Adaptive DNS attempts to take advantage of a domain’s entire response by predicting future queries and precomputing responses for them. When a user sends a new request, Adaptive DNS checks whether its response is already computed or still needs to be recomputed. If it is still needed, Adaptive DNS computes it on-demand but caches the result for future use with high efficiency by leveraging information from past requests. This allows us to save up to 80% of all bytes transferred over IPv4 in peak hours . The savings are achieved without requiring any change at the server side and while correctly servicing all existing DNS requests.
Adaptive DNS achieves these high savings by using three ingredients:
(1) predicting future queries based on past requests,
(2) precomputing responses for such lookups and
(3) caching such responses with high efficiency to save bandwidth while correctly servicing all existing requests.
How does it help?
It reduces network traffic caused due to frequent dns requests . Let’s say you are visiting google , your computer performs a DNS lookup before it can communicate with Google’s servers using this numerical address . Normally there is a 100ms delay in responding from the Server(google) side . This slows down web browsers and other applications which perform frequent domain lookups as they wait for the reply from the DNS server which can easily take 100ms or more .
How does it work?
Our proposal speeds up the web browser while reducing its memory consumption. To reduce the web’s latency, our system stores recently accessed DNS information in a cache on the web browser side. Then, when other hosts are requested to be resolved, Adaptive DNS first consults this cache before sending out any queries. Because of space limitations within mobile devices, users should not store all known hostnames in their cache nor can they learn new domains online because of reduced network connectivity or data transfer cost. Instead, our system adapts to the user’s browsing pattern by learning frequently visited domains offline and storing them persistently across sessions . Domain adaptation is achieved by using fine-grained domain labels
Benefits of internet Scaling:
1. Faster Browsing by performing fewer DNS lookups.
2. Save network bandwidth, money and time.
3. Reduce power consumption in mobile devices due to less processing involved in DNS lookups .
4. Lesser Network Traffic
5. Contributions to end-system security
6. Reduced power & memory consumption on mobile devices
Conclusion:
DNS can be enhanced to improve the performance of web browsers and other Internet applications. Our approach reduces end-to-end latency by up to 80% while reducing memory consumption by up to 80% and bandwidth consumption by up to 80% . This is a significant improvement over existing proposals, which focus on reducing memory consumption of DNS lookups at the expense of more frequent lookups. We believe that enhanced DNS will play a key role in improving overall Internet performance and saving costs for operators. Our future work includes optimizing Adaptive DNS for real-world use on client platforms such as smartphones.
