HenryR

11 ms·

61.
by HenryR 16y ago
Depends on the protocol, in general. Paxos is, fundamentally, a quorum-based system that deals with reordering of messages. It sacrifices liveness for correctness - if the proposer does not hear back from a majority of nodes (in the case of
62.
by HenryR 16y ago
This blog post that I wrote a few months ago also explains the same issue, and may be of interest for those looking for a separate explanation: http://www.cloudera.com/blog/2010/04/cap-confusion-problems-...
63.
by HenryR 16y ago
No, it's exactly the same. Availability is a guarantee that all requests are eventually responded to within some time bound, whatever that is. During the partition, availability is violated. Therefore it's not a CA system, but a C system.
64.
by HenryR 16y ago
In asynchronous networks it is surprisingly hard to detect failures, even of yourself. Reporting an error condition counts as an availability violation.
65.
by HenryR 16y ago
I'm afraid you're not quite correct. CAP says that, in the face of potentially arbitrary network partitions (which are precisely modelled by a set of dropped messages) you can be 100% consistent, or 100% available, but you can't be both. If
66.
by HenryR 16y ago
No, dropped packets are partitions. They really are. A partitionable network is modelled as one which may fail to deliver any subset of sent messages between nodes. The Gilbert and Lynch paper makes this explicit. The consistency guarante
67.
by HenryR 16y ago
Cloudera is hiring: http://www.cloudera.com/company/careers/ UX, UI, PM, distributed systems engineer, operations engineer and more. We're in the Bay Area, down in Palo Alto and are genuinely a great company for which to work. If you're i
68.
by HenryR 16y ago
Step 1: Don't use threads for parallelism Threads are a reasonable structuring tool for expressing concurrency, which is useful for laying out your code in a maintainable, easy-to-reason-about way. Mainstream Python implementations have c
69.
by HenryR 16y ago
You're correct in a sense - virtual memory gives each process its own completely independent memory space that's addressed linearly. In order to preserve the illusion of independence, the OS has to deal with the possibility that the sum of
70.
by HenryR 16y ago
You can guarantee consistency; but what you must then lose is the ability to make progress in all situations ('liveness'). A trivial commit protocol that is consistent but not live simply sends no messages. All updates that succeed are cons
71.
Chuck Norris Hudson Plugin(wiki.hudson-ci.org)
1 points by HenryR 17y ago|0 comments
72.
by HenryR 17y ago
Yes, this is essentially the same idea. The problem with Lamport clocks is that every two pairs of events can be ordered - since a < b or b < a or a = b for all integers a and b. This means that two events that are not causally re
73.
by HenryR 17y ago
Computer science is the formal study of problem solving. Hence, algorithms - deterministic, mathematically tractable, general solutions to problems. Hence programming language theory - ways to efficiently express solutions to problems. Henc
74.
by HenryR 17y ago
No - all I know is based on rumour.
75.
by HenryR 17y ago
I don't agree with all his conclusions. In particular, it seems as though he is saying 'eventual consistency is no practical good' and then beating Dynamo for a few paragraphs with the same stick. The most often quoted example of Dynamo's u
76.
by HenryR 17y ago
I think this is a slightly more detailed version: http://www.enterpriseintegrationpatterns.com/docs/IEEE_Softw...
77.
by HenryR 17y ago
Although this paper didn't know it at the time, this was one of the first statements of what would eventually be called the Byzantine Generals Problem (there is a paper of that name by the same authors which has similar content, just expand
78.
by HenryR 17y ago
http://www.siggraph.org/ - this year's conference is pretty soon, I think.
79.
by HenryR 17y ago
Pretty flawed, unfortunately. I left this as a comment, but it has yet to be moderated through. ---- You’re doing a good thing by trying to elucidate basic concepts, but I’m afraid your article has a bunch of errors that means it’s not as h
80.
by HenryR 17y ago
Hi - author here. Yes, you're correct that I actually don't get as far as the computation. That's coming next - at over 2,000 words it became apparent that the series needed to be segmented. (I should have called this part 0, or maybe part
81.
Barbara Liskov’s Turing Award, and Byzantine Fault Tolerance (hnr.dnsalias.net)
6 points by HenryR 18y ago|0 comments
82.
by HenryR 18y ago
If you know of any other high quality, advanced undergraduate or graduate level courses in these areas, please leave them in the comments and I'll update the list.
83.
Links to advanced computer science courses - systems, algorithms and mathematics.(hnr.dnsalias.net)
21 points by HenryR 18y ago|3 comments
84.
Paper review from OSDI '08: FlightPath, a BAR tolerant peer to peer streaming protocol(hnr.dnsalias.net)
3 points by HenryR 18y ago|0 comments
85.
A tutorial on the Paxos consensus protocol(hnr.dnsalias.net)
2 points by HenryR 18y ago|0 comments
86.
A walkthrough guide to Google's BigTable(hnr.dnsalias.net)
3 points by HenryR 18y ago|0 comments
87.
Overview of PNUTS - Yahoo's take on distributed data storage a la Dynamo (link fixed)(hnr.dnsalias.net)
1 points by HenryR 18y ago|0 comments
88.
A short tutorial on Binomial Heaps with Python(hnr.dnsalias.net)
4 points by HenryR 18y ago|0 comments
89.
The Google File System - Notes and Commentary(hnr.dnsalias.net)
3 points by HenryR 18y ago|0 comments
90.
Chrome: The Real GoogleOS?(hnr.dnsalias.net)
1 points by HenryR 18y ago|0 comments

More ›