Tuesday, August 12, 2025

@scale networking event to learn from Meta networking researchers

One of my goal in 2025 is to get back into the swing of things, which is get back closure to hardware and stay in research mode. I used to attend many social events in my passionate field "Internetworking" and share my perspective. I've been on a long hibernation for the past 6 years. Hoping to get back to Networking research in year 2025. Aug-13th @scale networking gives me a perfect restart. Through my DataCenter networking social circle i got invited to a much anticipated networking related event by Meta @scale networking. I plan to dedicate my entire day to observe and learn as much as possible. 

What are my expectations, 

  1. Are there any new topology level innovations happened at Meta? What did they learn from different network topologies used to carry out AI work-loads? What is new after 5-stage + planes, Dragonfly and dragonfly+ topologies
  2. How Meta identifies potential networking side bottleneck from application level metrics?
  3. How meta identifies regular system level problems and root-causes them?
  4. What Meta does to stay ahead of ongoing networking innovations?
Networking plays crucial role in getting AI faster and better. Networking product and services grew exponentially in recent years especially recent announcements on TH6 ( Tomahawk6 ) from Broadcom has some unheard scales such as,

  • Number of ports per SoC ( System on chip) - 1024 channelized port in 1x chip.
  • Per port bandwidth and collectively per chip bandwidth. Popularly called as TH6 ( Tomahawk 6) expected to deliver 102.4 Terabits per second. 
  • Supports both scale-up and scale-out network
  • chiplet architecture
  • UEC support
P.S> Thanks to my 10-year old son Advik for proof reading.

Wednesday, July 18, 2018

Interesting Meetup Today - Mark Burgess - Trust and Microservices

IT, cloud and Virtualization going through transformations on their own. Compute infra went beyond virtualization, container are in main stream. Administering complex systems made-up of micro services  without compromising security is critical. Simplifying their interaction is also essential.. I've many thoughts around this area and working in solving some of the problems.

Learn and share is the best way to tune your ideas and meetup helps you to do it. I'm looking forward to shape up my thoughts, learn, share and listen from others. Lead speaker has very promising profile, I'm excited to take part in this meetup. Looking froward to meet fellow engineers to learn and explore.


Meetup link: https://www.meetup.com/SVDevOps/events/252034643/

Speaker profile:

About Mark Burgess

Mark Burgess is a theoretician and practitioner in the area of information systems, whose work has focused largely on distributed information infrastructure. He is known particularly for his work on Configuration Management and Promise Theory. He was the principal Founder of CFEngine, and is emeritus professor of Network and System Administration from Oslo University College. He is the author of numerous books, articles, and papers on topics from physics, Network and System Administration, to fiction. He also writes a blog on issues of science and IT industry concerns.

https://www.linkedin.com/in/markburgessoslo/

Saturday, June 16, 2018

Why Networking Industry is very relevant to stay in current Technology Transformation

Happy to start my writing after Fathers day week end. My 3yr old presented me with a beautiful type-writer show piece, His message says Dad get back to your writing and share your you learning. Inspiration has no boundary and limit, This passage marks a new beginning in my writing. I often get this question "How to become a successful Networking Test Engineer ?", I will try to answer this questions with my living experience as Networking Test Engineer.

Rather than giving bullet points and cheat sheet style advice, I prefer intuitive self motivated career grown, Hope you have the motivation and patience to travel this with me. Lets begin our conversation with the question WHY?

Step#1 Why Networking Industry and Why Now?

You cannot deny the fact "Internet is everything". Computer networking made it possible to connect all of us. Number of connected devices are going up in an unimaginable level. As a history goes a 231 node comprised ARAPNET in 1981 become billions of connected devices, one single family has more than 4 connected devices. That is the power of connectivity technologies. I feel proud to contribute to this technology transformation, as I see we have lot more problems to solve and innovate. IMO, Networking industry played a critical roles in current innovation era. If you are looking to find a best domain to start your career Networking Industry is the best Industry and Now is the best time.

Why Now, Let me highlight few points from Cisco Visual Networking Index data,

Almost half a billion (429 million) mobile devices and connections were added in 2016
Mobile network connection speeds will increase threefold by 2021
More than three-fourths of the world’s mobile data traffic will be video by 2021

Data service changed from an optional expensive add-on to default service to every services. The data we generate, share and store is multifold. Networking as a core technology which includes Connectivity technologies like 2/3/4G LTE and upcoming 5G, WiFi/WiMax, High-Frequency radios and Fiber optics operate with different principle. These technologies created product vendors (One who makes a product based on a given Technology), Service Providers (One who deploys the product to offer a service and consumers. A product suppose to function with an agreed methods called protocols so that a base-station/Service-provider installed products can talk to a consumer devices.

Given the multifold nature of the industry available opportunities are enormous. There is a space for every one.Having a right skill will get you the right place with good rewards. Learning the required is important

Tuesday, June 13, 2017

Data Analytics journey in Networking - An optional to Essential

Data Analytics and Machine Learning, Two of the most popular buzz words in Tech industry not only entered Networking industry did make a bigger impact. Gone are the days where networking equipment operators thrust on syslog servers and SNMP based monitoring tools to analyze and manage networks. Modern networks need agile ways to monitor networks. Super intelligent traffic engineering methods intrduced by  modern networking gears forwarding techniques offer many different seamless failover technique, age old monitoring utilities has less chance of detecting such fail-overs. Also, operators view of monitoring tools as CAPEX had a hard time in integrating monitoring utilities with usage billing. Netflow and sFlow as a flow analysis method or flow monitoring method couldn't tale off due to inherent complexity is building external data store and retrieve layer. Network monitoring utilities failed to stitch networking events with application/server performance, hence ended up in the usual suspect list on every outage.

Networking Industry prayed for a New King with abundant power and easy to adapt integration techniques. From the ray of hope Network Data Analytics raised to industries prayer. Data Analytics had become an essential part of Networking. It attacks traditional network monitoring bottle necks with cognitive echo system.  Echo system consists of Tailor made programmable asics and easy to use scalable analytics core engine with army of API support. 


  • Programmable ASIC standout by line rate analytics data export. Cisco, Broadcom (Avago), Intel had come up with stream analytics directly from ASIC. similar to constructing Data traffic, HW analytics tables will get stream out of switch/router.
  • Network security, a growing vertical in networking eco system relies on collected pieces of information such as data signature and flow patterns. Storing time-sensitive mass data is critical for Network security. Streaming Telemetry makes it easy by means of less overhead in consuming high volume of data. 
  • Traditionally SNMP implemented as an application to listen to asic exports. SNMP responded to external entities request,  this method is a big overhead in networking gears CPU performance. Streaming telemetry removes these bottle necks. 
  • Analytics engine need not be a new software. Regular business data analytics engines with a protocol buffer to decode streamed data is good enough. 

Who is playing a key role in Network Data Analytics? 
Cisco as an Industry leader has fantastic solutions. 
Tetration analytics -  real-time data center analytics, combines network and server sensors to simplify SDN, migration to the cloud, and white-list security.
LancopeLancope provides network behavior analytics, threat visibility and security intelligence to help protect companies against top cyber security threats.

Splunk - with host of Network analytics centric tools. Splunk is a bug player in Network Data Analytics
IllumioIllumination provides a live application dependency map across your environments showing workloads, applications, and traffic flows so you can see how applications communicate and identify violations quickly.

Network security companies - PaloAlto Networks, fire eye have their own Network data analytics utilities.

P.S. Above list represent only a small set which i am aware of.

Saturday, September 3, 2016

DATA is the Clear winner Finally an Indian Telco wakes up

Reliance Industries chairman Mr. Mukesh Ambani has announced tariff plans of much anticipated Reliance Jio (4G+ network). Customer billing is only in terms of the amount of Data used per month Voice calls are completely free.

“For Reliance... data is the new oil, and intelligent data is the new petrol”Mukesh Ambani, Chairman and MD – Reliance Industries.

 I am sure this move is about to create a big dent on other operators revenue and business plan. If you have any doubt on Reliance Jio's plan and it's business objective you landed in correct page. Let's bisect read the fine print,

Approximately 7 years ago I wrote an article about Data volume in a mobile network and its importance in capacity planning. I also mentioned about the need for an innovative business plan to harvest the effective efficiency of Data and hence increase revenue. 

Most telecom industry experts know the importance of ARPU - Average Revenue Per User. This magic word helps to forecast revenue and do capacity planning. RF architect may not directly agree with me but our common variable erlang assists to determine ARPU, which is a clear business term for ROI calculation. 

   Erlang is defined as an amount of voice traffic in an hour. .i.e Total voice minutes / 60. Total voice minutes = no.of users x Avg usage per user.

Long ago Operators have moved away from erlang. simply because Voice channel is not the only one carried by signal Data is carried together. In recent times more DATA channels are used than Voice. Erlang is history and hence should be Voice based billing

Data usage is simply growing at an exponential rate than Voice. A photo you clicked and shared, A txt you received/forwarded and every call/emoji you use in watsapp/imessage/hangout/G+ simply Data. 

Read following data points from recent article from frost, 
  • An average 4G LTE smartphone consumes upwards of 2.5 GB of cellular data every month
  • Connected tablets are being used more and generating approximately 2.0 GB of cellular data on 4G networks
  • Connected laptops and air cards (yes these still exist!) consume more than 5 GB of cellular data every month
One more data point about DATA traffic growth from Internetworking product leader CISCO,

Sorry for fooling you with slightly incorrect detail. I said SLIGHTLY remember i am building my data points for DATA. The actual truth is every voice call you make is transferred as Data, we live in 4G world your voice is actually carried as Data in Teleco's core network. Billing you by per minute is simply a way to fool you. There is nothing like Roaming, A voice call is IP datagram gets routed across state (say CA to Tx or TN to UP) mostly similar to Data (From sender (your cell) to Server and server to Receiver). So, from a Telecom infrastructure perspective, you need to build and maintain same intelligence for Data and Voice by allowing a not growing segment Reliance is marketing well to capture customers from a competitor. 

Bill your customer based on DATA usage is the way forward. Reliance Jio is making a bold move and a logical move. Good luck Jio, i am sure Reliance knows to win a customer you need to do lot more than an innovative billing. Quality metrics are very important than simplified billing. 

All the best RELIANCE JIO. 



Thursday, July 28, 2016

Is this a right time to buy a home in Bay area?

I really don't want to steer away from my blog focal point, which in Networking technologies. This topic is most discussed in office and in many social gathering. I would like to record my 2 cents with this post,

DISCLAIMER: I am not an economist OR Banker OR Real-estate agent. This is not my answer to a Quora question.

US house price goes through cycle your purchase point the cycle decides your yield. IMO, at this moment [July, 2016] ]US economy is not ready for another recession. Do not expect 30-50% correction in home prices. Economy mends (oct-dec 2013 house market index corrected, may-july 2016 index corrected), It's really hard to find these drop points. Given the housing demand in bay area, It takes about 3-5 failed offer to get one successful accepted offer. 
House rents have gone up disproportionately with individual income. 2BHK apartment, which costed $1550 a month in 2010 is at $2550 today. Give Interest rates is still at an attractive rate. For the same floor plan mortgage is less than rent, since earning on cash is 1% best case. 

Few facts to substantiate my view on housing bubble timing, 

Recovery time (Down trend to the beginning of appreciation) gives a good hint about Market explode time (continuous UP trend to down). 
  • 2008 recession affected housing market by 60% (CA value, not national) 
  • Other corrections .com (2000), underwriting collapse (199x) impacted only by 10%
  • 2007 to 2011 => Down trend, we are currently in up trend 2011 to 2016. 3.5 yrs Vs 5 years
Factors affecting house market, 
  • Bubble needs Irrational fraudulent and excess greed – IMO, Three factors can make this happen: Tech valuations irrational, Post Brexit and politics
    • Effect of TECH valuations footprint is limited to select pockets
  • FED support to housing market. Owning a house is relatively cheaper than renting one.
  • 5 years of UP trend is too short for a bubble to form. 

Find a value for money place. Look at your need, you don’t need high school rating of 10 with elementary starting kid.

Reference sites,

Tuesday, January 12, 2016

Quantifying traffic policer rate with Burst-Size value - test method and calculation

Quality Of Service (QoS) dictates how a packet or flow is handled in networking world. Task is not as simple as the sound of this magic three letter word. Many functionalities of QOS work together to accomplishes this trivial task.  QoS complexity is dictated by number of modifications performed to return intended result, QOS  concepts are hard to perceive for many networking professionals.

Let's add one of QoS functionality "POLICING/RATE-LIMITING"

Traffic policing is one of the commonly used QoS feature. Policer/rate-limiter helps to allow only defined packet rate for interested flow. Various other sub-tasks like set queue.no or mark  a specific bit in packet for violated and conformed actions are also possible.

Typical traffic policing configuration looks similar to.
 Police cir 100 mbps bc 200ms pir 200 mbs bc 20 ms conform transmit exceed set-dscp 3 violated drop

above config allows traffic rate  of 100kbs, sets dscp value 3 for flow between 100mbps to 200 mbps and drop any further traffic.

Most networking engineers don't really know that traffic is not actually policed at 100 mbbs, there is more to it, BC helps to define it.

What does "BC - Burst Count" do?
Why my actual policer rate is more than defined rate?
How does it helps on practical traffic flows?

Let's find answer here.

Burst count helps to adjust policer rate to absorb traffic burst. Real world traffic flow is bursty in nature. Handling burst help to have handle on policer rate without increasing the traffic drop point.
 very rare to see a constant rate of data flow, even if you observe high traffic rate, it is definitely constituted by several mice flows than one giant elephant flow.

BC in time representation translates to Bytes based on port speed.

for a 100mbps policer rate  200 ms BC
Policer bandwidth in Bytes is,

(100 x  1000 x 1000 )  bps x 0.2 Sec  = 2500 KBytes
--------------------------------------------
  8 bits in a byte

On a 10gbps port speed this translates  (2500 KB / 10gbps)to, 2msec of burst duration.

Bandwidth rate in 2msec on 100mbps link becomes 0.2mbps. Effective policer rate is 100.2 Mbps


Typical BC metrics are either in time (ms/micro-sec/sec) or in Bytes. Now that you know the conversion. effective policer rate can be easily determined.