Decade Inward Review
We are entering 2020, as well as this is a goodness fourth dimension to retrospect as well as review the yesteryear decade from the lens of this blog, which started belatedly 2010.
Spanner's utilisation of synchronized clocks inwards distributed databases was a large milestone. This piece of job led us to intend most hybrid vector clocks, as well as subsequently hybrid logical clocks. This subsequently led us to prepare timely protocols, inwards continuation of this line.
Cloud computing has e'er been an exploratory technical newspaper called stabilization inwards the cloud. I intend that is all the same an opened upwards as well as interesting problem. Recently serverless (function equally a service) is all the rage inwards the cloud environments as well as at that topographic point bring been several interesting papers on the topic.
I actually got into Paxos inwards this lastly decade. I didn't facial expression to autumn for Paxos protocols this hard. I had outset encountered Paxos to a greater extent than or less 1999 inwards the distributed systems reading seminars I attended equally a PhD student. I gauge I had liked it dorsum then, exactly it didn't brand a large impression on me. From 2007-2015 Paxos gradually became to a greater extent than as well as to a greater extent than pop as well as of import inwards cloud computing. In this weblog nosotros had more than twenty Paxos posts alone inwards the lastly 2 years.
My students as well as I bring a love-hate human relationship alongside Paxos. We empathise it rattling good as well as are amidst the superlative experts on the topic. But, unfortunately, academia lost some of its excitement most Paxos as well as a therefore called "Paxos fatigue" developed. Because of this mutual frigidity shoulder, nosotros enjoin ourselves that perchance nosotros should live working on other things. On the other hand, Paxos is all the same transforming as well as ruling over large scale distributed systems deployments inwards the manufacture (one recent illustration is a vast unexplored algorithm pattern space, as well as to a greater extent than piece of job is needed to tailor Paxos for specific distributed systems deployments, topologies, as well as workloads. A striking bear witness of this arrived inwards 2017 alongside the flexible quorum breakthrough which came unanticipated almost thirty years after the Paxos protocol was outset proposed. This farther opened upwards the pattern infinite for customizing Paxos to unlike environments as well as workloads, which is yet to live fully realized. We promise nosotros volition live able to convince to a greater extent than researchers to assist most as well as piece of job on these problems. In whatever case, nosotros can't look to delineate ourselves away from working on Paxos variants. They are a lot of fun. I am working on Paxos Unpacked now.
Ok, plenty most Paxos...
In 2016, the machine learning champaign exploded as well as went mainstream. Distributed systems back upwards for machine learning became a hot topic. I learned some machine learning yesteryear next online courses as well as reading papers. I actually appreciated the great mathematics, differential computation employed here. As far equally developing some surveys, as well as idea most the theme for a while. But I gave upwards after sometime, thinking that it would live difficult to produce principled algorithmic piece of job hither because the ML champaign industrial plant rattling closed to the application, as well as the solutions are pretty application-specific. It turns out that I was judgmental, as several overnice algorithmic as well as distributed systems started coming out recently.
Blockchains got a lot of hype recently. It took me a long fourth dimension to acquire inwards to a seminar on blockchains inwards Spring 2018, I started appreciating some of the goodness piece of job done inwards this domain, as well as for certain parts of the vision for decentralized computing. I dearest the premise of ICOs for democratizing the stock marketplace position as well as of smartcontracts for enabling decentralized e-trade without whatever middleman (please hurry up, nosotros remove a decentralized search engine as well as large information analytics for this to work). Unfortunately the blockhain champaign has been perpetually overhyped as well as this damages the progress inwards the field. I intend nosotros finally started to consider the hype dying as well as to a greater extent than corporation piece of job inwards the champaign reconvening.
Bitcoin was released inwards 2009 as well as most of us missed it equally nosotros come inwards 2010. What are some trends that are brewing silently that nosotros are missing now?
IOT as well as 3D printing areas bring been seeing increasingly to a greater extent than interest. But at that topographic point has non been whatever revolutionary breakthrough yet equally far equally I tin see.
Differentiable programming seems interesting to me.
Quantum computing is also seeing some overnice progress, exactly I don't know much most the area.
The cloud/datacenter computing came a long means inwards this lastly decade. I intend the serverless model, dataflow architectures for analytics/transactions, as well as the utilisation of RDMAs are some promising trends. I am also happy to consider increased adoption of formal methods as well as verification for distributed systems. But, maybe since I am embedded equally good much inwards the field, I am unable to position a large striking clearly for the side yesteryear side coming decade.
I started posting regularly on September 2010. I wanted to acquire into the cloud computing domain, therefore I needed to accumulate background on cloud computing work. I decided that equally I read papers on cloud computing, I volition post service a summary to this blog. I idea if I could explicate what I learned from the papers inwards my ain words, I would internalize those lessons better. And if others read those summaries as well as benefit, that is an extra plus.Initially I reviewed as well as posted newspaper summaries on large information processing systems as well as NoSQL distributed databases to grab upwards on these areas. Around 2010s, misrepresentation of the CAP theorem yesteryear NoSQL proponents was a problematic issue. I covered some papers that try to clarify this issue. Throughout the years, I covered many papers that discussed the unlike I was at Microsoft Azure Cosmos DB for my sabbatical as well as got to larn to a greater extent than most this domain.
Spanner's utilisation of synchronized clocks inwards distributed databases was a large milestone. This piece of job led us to intend most hybrid vector clocks, as well as subsequently hybrid logical clocks. This subsequently led us to prepare timely protocols, inwards continuation of this line.
Cloud computing has e'er been an exploratory technical newspaper called stabilization inwards the cloud. I intend that is all the same an opened upwards as well as interesting problem. Recently serverless (function equally a service) is all the rage inwards the cloud environments as well as at that topographic point bring been several interesting papers on the topic.
I actually got into Paxos inwards this lastly decade. I didn't facial expression to autumn for Paxos protocols this hard. I had outset encountered Paxos to a greater extent than or less 1999 inwards the distributed systems reading seminars I attended equally a PhD student. I gauge I had liked it dorsum then, exactly it didn't brand a large impression on me. From 2007-2015 Paxos gradually became to a greater extent than as well as to a greater extent than pop as well as of import inwards cloud computing. In this weblog nosotros had more than twenty Paxos posts alone inwards the lastly 2 years.
My students as well as I bring a love-hate human relationship alongside Paxos. We empathise it rattling good as well as are amidst the superlative experts on the topic. But, unfortunately, academia lost some of its excitement most Paxos as well as a therefore called "Paxos fatigue" developed. Because of this mutual frigidity shoulder, nosotros enjoin ourselves that perchance nosotros should live working on other things. On the other hand, Paxos is all the same transforming as well as ruling over large scale distributed systems deployments inwards the manufacture (one recent illustration is a vast unexplored algorithm pattern space, as well as to a greater extent than piece of job is needed to tailor Paxos for specific distributed systems deployments, topologies, as well as workloads. A striking bear witness of this arrived inwards 2017 alongside the flexible quorum breakthrough which came unanticipated almost thirty years after the Paxos protocol was outset proposed. This farther opened upwards the pattern infinite for customizing Paxos to unlike environments as well as workloads, which is yet to live fully realized. We promise nosotros volition live able to convince to a greater extent than researchers to assist most as well as piece of job on these problems. In whatever case, nosotros can't look to delineate ourselves away from working on Paxos variants. They are a lot of fun. I am working on Paxos Unpacked now.
Ok, plenty most Paxos...
In 2016, the machine learning champaign exploded as well as went mainstream. Distributed systems back upwards for machine learning became a hot topic. I learned some machine learning yesteryear next online courses as well as reading papers. I actually appreciated the great mathematics, differential computation employed here. As far equally developing some surveys, as well as idea most the theme for a while. But I gave upwards after sometime, thinking that it would live difficult to produce principled algorithmic piece of job hither because the ML champaign industrial plant rattling closed to the application, as well as the solutions are pretty application-specific. It turns out that I was judgmental, as several overnice algorithmic as well as distributed systems started coming out recently.
Blockchains got a lot of hype recently. It took me a long fourth dimension to acquire inwards to a seminar on blockchains inwards Spring 2018, I started appreciating some of the goodness piece of job done inwards this domain, as well as for certain parts of the vision for decentralized computing. I dearest the premise of ICOs for democratizing the stock marketplace position as well as of smartcontracts for enabling decentralized e-trade without whatever middleman (please hurry up, nosotros remove a decentralized search engine as well as large information analytics for this to work). Unfortunately the blockhain champaign has been perpetually overhyped as well as this damages the progress inwards the field. I intend nosotros finally started to consider the hype dying as well as to a greater extent than corporation piece of job inwards the champaign reconvening.
MAD questions
What trends are brewing that nosotros are missing?Bitcoin was released inwards 2009 as well as most of us missed it equally nosotros come inwards 2010. What are some trends that are brewing silently that nosotros are missing now?
IOT as well as 3D printing areas bring been seeing increasingly to a greater extent than interest. But at that topographic point has non been whatever revolutionary breakthrough yet equally far equally I tin see.
Differentiable programming seems interesting to me.
Quantum computing is also seeing some overnice progress, exactly I don't know much most the area.
The cloud/datacenter computing came a long means inwards this lastly decade. I intend the serverless model, dataflow architectures for analytics/transactions, as well as the utilisation of RDMAs are some promising trends. I am also happy to consider increased adoption of formal methods as well as verification for distributed systems. But, maybe since I am embedded equally good much inwards the field, I am unable to position a large striking clearly for the side yesteryear side coming decade.
0 Response to "Decade Inward Review"
Post a Comment