Distributed Hash Tables, Part I
Listing 3. update.py
def update(node): for x in range(k): oldEntry=node.finger[x] node.finger[x]=findNode(oldEntry, (node.id+(2**x)) % (2**k))
Listing 4. finger-lookup.py
def findFinger(node, key): current=node for x in range(k): if distance(current.id, key) > \ distance(node.finger[x].id, key): current=node.finger[x] return current def lookup(start, key): current=findFinger(start, key) next=findFinger(current, key) while distance(current.id, key) > \ distance(next.id, key): current=next next=findFinger(current, key) return current
So far we have more or less defined the original version of the Chord DHT design as it was described by the MIT team that invented it. This is only the tip of the iceberg in the world of DHTs, though. Many modifications can be made to establish different properties from the ones described in the original Chord paper, without losing the logarithmic performance and guaranteed lookups that Chord provides.
One property that might be useful for a DHT is the ability to update the finger table passively, requiring periodic lookups to be done in order to refresh the table. With MIT Chord, you must do a lookup, hitting O(log n) nodes for all k items in the finger table, which can result in a considerable amount of traffic. It would be advantageous if a node could add other nodes to its finger table when they contacted it for lookups. As a conversation already has been established in order to do the lookup, there is little added overhead in checking to see if the node doing the lookup is a good candidate for the local finger table. Unfortunately, finger table links in Chord are unidirectional because the distance metric is not symmetrical. A node generally is not going to be in the finger tables of the nodes in its finger table.
A solution to this problem is to replace Chord's modular addition distance metric with one based on XOR. The distance between two nodes, A and B, is defined as the XOR of the node IDs interpreted as the binary representation of an unsigned integer (Listing 5). XOR makes a delightful distance metric because it is symmetric. Because distance(A, B) == distance(B, A), for any two nodes, if A is in B's finger table then B is in A's finger table. This means nodes can update their finger tables by recording the addresses of nodes that query them, reducing significantly the amount of node update traffic. It also simplifies coding a DHT application, because you don't need to keep a separate thread to call the update method periodically. Instead, you simply update whenever the lookup method is called.
Listing 5. xor-distance.py
def distance(a, b): return a^b # In Python, this means a XOR b, # not a to the power of b.
An issue with the design presented so far is the paths to a given node are fragile. If any node in the path refuses to cooperate, the lookup is stuck. Between any two nodes there is exactly one path, so routing around broken nodes is impossible. The Kademlia DHT solves this by widening the finger table to contain a bucket of j references for each finger table slot instead of only one, where j is defined globally for the whole network. Now j different choices are available for each hop, so there are somewhere around j*log(n) possible paths. There are less than that, though, paths converge as they get closer to the target. But, the number of possible paths probably is greater than 1, so this is an improvement.
Kademlia goes further and orders the nodes in the bucket in terms of recorded uptime. Older nodes are given preference for queries, and new references are added only if there are not enough old nodes. Besides the increased reliability of queries, this approach offers the added benefit that an attack on the network in which new nodes are created rapidly in order to push out good nodes will fail—it won't even be noticeable.
It's important to understand that these different properties are not tied to a particular DHT implementation. We gradually have built up a DHT design from scratch, developed it into something that resembles Chord, then modified it to be more like Kademlia. The different approaches can be more or less mixed and matched. Your finger table buckets can have 1 slot or j slots, depending on whether you use modular addition or XOR for your distance metric. You can always follow the closest node, or you can rank them according to uptime or according to some other criteria. You can draw from several other DHT designs, such as Pastry, OceanStore and Coral. You also can use your own ideas to devise the perfect design for your needs. Myself, I have concocted several modifications to a base Chord design to add properties such as anonymity, Byzantine fault-tolerant lookups, geographic routing and the efficient broadcasting of messages to enter the network. It's fun to do and easier than you think.
Now that you know how to create your own DHT implementations, you're probably wondering what kind of crazy things you can do with this code. Although there probably are many applications for DHTs that I haven't thought of yet, I know people already are working on such projects as file sharing, creating a shared hard drive for backing up data, replacing DNS with a peer-to-peer name resolution system, scalable chat and serverless gaming.
For this article, I've tied the code together into a fun little example application that might be of interest to readers who caught my interview on the Linux Journal Web site about peer-to-peer Superworms (see Resources). The application is a distributed port scanner that stores results in the simulated DHT (Listing 6). Given a fully functional DHT implementation, this script would have some handy properties. First, it allows multiple machines to contribute results to a massive scanning of the Internet. This way, all of the scanning activity can't be linked with a single machine. Additionally, it avoids redundant scanning. If the host already has been scanned, the results are fetched from the DHT, avoiding multiple scans. No central server is required to hold all of the results or to coordinate the activities of the participants. This application may seem somewhat insidious, but the point is it was trivial to write given the DHT library. The same approach can be used in other sorts of distributed projects.
Practical Task Scheduling Deployment
July 20, 2016 12:00 pm CDT
One of the best things about the UNIX environment (aside from being stable and efficient) is the vast array of software tools available to help you do your job. Traditionally, a UNIX tool does only one thing, but does that one thing very well. For example, grep is very easy to use and can search vast amounts of data quickly. The find tool can find a particular file or files based on all kinds of criteria. It's pretty easy to string these tools together to build even more powerful tools, such as a tool that finds all of the .log files in the /home directory and searches each one for a particular entry. This erector-set mentality allows UNIX system administrators to seem to always have the right tool for the job.
Cron traditionally has been considered another such a tool for job scheduling, but is it enough? This webinar considers that very question. The first part builds on a previous Geek Guide, Beyond Cron, and briefly describes how to know when it might be time to consider upgrading your job scheduling infrastructure. The second part presents an actual planning and implementation framework.
Join Linux Journal's Mike Diehl and Pat Cameron of Help Systems.
Free to Linux Journal readers.Register Now!
- Stunnel Security for Oracle
- Murat Yener and Onur Dundar's Expert Android Studio (Wrox)
- SourceClear Open
- SUSE LLC's SUSE Manager
- My +1 Sword of Productivity
- Managing Linux Using Puppet
- Google's SwiftShader Released
- Non-Linux FOSS: Caffeine!
- Parsing an RSS News Feed with a Bash Script
- SuperTuxKart 0.9.2 Released
With all the industry talk about the benefits of Linux on Power and all the performance advantages offered by its open architecture, you may be considering a move in that direction. If you are thinking about analytics, big data and cloud computing, you would be right to evaluate Power. The idea of using commodity x86 hardware and replacing it every three years is an outdated cost model. It doesn’t consider the total cost of ownership, and it doesn’t consider the advantage of real processing power, high-availability and multithreading like a demon.
This ebook takes a look at some of the practical applications of the Linux on Power platform and ways you might bring all the performance power of this open architecture to bear for your organization. There are no smoke and mirrors here—just hard, cold, empirical evidence provided by independent sources. I also consider some innovative ways Linux on Power will be used in the future.Get the Guide