DynaTrace's Top 10 Performance Problems taken from Zappos, Monster, Thomson and Co

DynaTrace in Top 10 Performance Problems taken from Zappos, Monster, Thomson and Co, has provided a useful compilation of performance problems, with potential solutions, that they've found while working with their clients. 

  1. Too Many Database Calls -  too many database query per request/transaction.
  2. Synchronized to Death - in a high-load or production environment over-synchronization results in severe performance and scalability problems.
  3. Too chatty on the remoting channels - too many calls across these remoting boundaries and in the end causes performance and scalability problems.
  4. Wrong usage of O/R-Mappers - incorrect usage of the framework itself too often results in unexpected performance and scalability problems within these frameworks.
  5. Memory Leaks - GC does not prevent memory leaks, it is important to release object references as soon as they are no longer needed.

Click to read more ...


Sponsored Post: VoltDB and Digg are Hiring

Who's Hiring?

VoltDB Field/Community Engineer

VoltDB is attracting more and more users every day. If you have a strong technical background in SQL and Linux, are experienced with production database deployments, and have a passion for customers and community, you could be just the person we are looking for.  Are you excited about the prospect of working with users to develop and deploy VoltDB applications, and about helping users participate in the thriving VoltDB community? If so, read on at their job page.

Get Your High Scalability Fix at Digg 

Interested in working on cutting-edge high-scale infrastructure at Digg? We're making a big investment in scaling and have committed to the NoSQL (Not only SQL) path with Cassandra. We're using other open-source infrastructure to help us scale including Hadoop, RabbitMQ, Zookeeper, Thrift, HDFS and Lucene. We're rewriting Digg from the ground up and we need amazing developers to join our world-class team. If you think you are up for the challenge, or you know someone who might be, take a look at our jobs page for more information.


DbShards Part Deux - The Internals

This is a follow up article by Cory Isaacson to the first article on DbShards, Product: dbShards - Share Nothing. Shard Everything, describing some of the details about how DbShards works on the inside.

The dbShards architecture is a true “shared nothing” implementation of Database Sharding. The high-level view of dbShards is shown here:

The above diagram shows how dbShards works for achieving massive database scalability across multiple database servers, using native DBMS engines and our dbShards components. The important components are:

Click to read more ...


Creating Scalable Digital Libraries

Like many other media content providers, libraries and museums are increasingly moving their content onto the Web.  While the move itself is no easy process (with digitization, web development, and training costs), being able to successfully deliver content to a wide audience is an ongoing concern, particularly for large libraries.

Much of the concern is financial, as most libraries do not have the internal budget or outside investors that for-profit businesses enjoy.  Even large university libraries will face serious budget constraints that even other university departments, such as science and technology would not face.

Creating a scalable infrastructure and also distributing a large digital collection that can handle multiple requests, requires planning that many librarians have not even imagined.  They must stop thinking in terms of "one-item-per-customer" and start thinking in terms of numerous users accessing the same information simultaneously.

Click to read more ...


So, Why is Twitter Really Not Using Cassandra to Store Tweets?

A firestorm of accusations circled around recently saying that Cassandra, the elected-by-major-adopters emperor of the NoSQL movement, has no clothes. It was said Twitter was dumping Cassandra; Reddit outages were linked to Cassandra; and even Facebook, Cassandra's cradle of birth, was said to have abandoned Cassandra. Shouts of NoSQL Fail! were heard in the streets. Much gloating followed. Is the emperor really naked? Casually dressed maybe, but not naked.

(Note: after this point the article contains a flow chart that is NSFW. Some people are very sensitive about cussing, so if that's you, please go back, don't read on. Danger! There are no nude pictures or anything, just some strong language. But this is my most favorite flow chart of all time, so it's worth it :-)

Is Twitter really abandoning Cassandra?

Click to read more ...


Hot Scalability Links for July 9, 2010

  • Facebook serves 3 billion Like buttons a day says VentureBeat.
  • CloudScaling reports: Rumor Mill: Google EC2 Competitor Coming in 2010? It looks like GAE for PaaS and an EC2 clone for IaaS.
  • Tweets of gold:
    • alandipert: scalability is a drug
    • seldo: Scalability lesson #23: if any part of your system involves a list that gets bigger over time, eventually that list will become too big.
    • obfuscurityHer: "Go look at the pictures on the database." Me: "You mean our fileserver?" Her: "Whatever." 
    • luiscab: Ouch, I just read on an Info Mgmt rag that Hadoop could easily be an acronym for "Heck, Another Darn Obscure Open-source Project."
    • sanity: Depressed about how much time I've had to spend searching for the right database solution for a new project. Each has it's flaws
    • ioshints: You cannot take a car, grow it 10 times and expect to get a mining truck. 

Click to read more ...


Cloud AWS Infrastructure vs. Physical Infrastructure

This is a guest post by Frédéric Faure (architect at Ysance) on the differences between using a cloud infrastructure and building your own. Frédéric was kind enough to translate the original French version of this article into English.

I’ve been noticing many questions about the differences inherent in choosing between a Cloud infrastructure such as AWS (Amazon Web Services) and a traditional physical infrastructure. Firstly, there are a certain number of preconceived notions on this subject that I will attempt to decode for you. Then, it must be understood that each infrastructure has its advantages and disadvantages: a Cloud-type infrastructure does not necessarily fulfill your requirements in every case, however, it can satisfy some of them by optimizing or facilitating the features offered by a traditional physical infrastructure. I will therefore demonstrate the differences between the two that I have noticed, in order to help you make up your own mind.

Click to read more ...


Strategy: Recompute Instead of Remember Big Data

Professor Lance Fortnow, in his blog post Drowning in Data, says complexity has taught him this lesson: When storage is expensive, it is cheaper to recompute what you've already computed. And that's the world we now live in: Storage is pretty cheap but data acquisition and computation are even cheaper.

Jouni, one of the commenters, thinks the opposite is true: storage is cheap, but computation is expensive. When you are dealing with massive data, the size of the data set is very often determined by the amount of computing power available for a certain price. With such data, a linear-time algorithm takes O(1) seconds to finish, while a quadratic-time algorithm requires O(n) seconds. But as computing power increases exponentially over time, the quadratic algorithm gets exponentially slower.

For me it's not a matter of which is true, both positions can be true, but what's interesting is to think that storage and computation are in some cases fungible. Your architecture can decide which tradeoffs to make based on the cost of resources and the nature of your data. I'm not sure, but this seems like a new degree of freedom in the design space.


Hot Scalability Links for July 2, 2010

  • What says 4th of July like Nathan's ultimate scalable hot dog eating contest? This totally requires a scale-up strategy.
  • Facebook at 60,000 servers and counting.
  • Deepak Singh has collected some impressive massive data stats on extreme Hadoop usage: Facebook: 36 PB of uncompressed data, 2250 machines, 23,000 cores, 32 GB of RAM per machine, processing 80-90TB/day; Yahoo: 70 PB of data in HDFS, 170 PB spread across the globe, 34000 servers, Processing 3 PB per day, 120 TB flow through Hadoop every day; Twitter: 7 TB/day into HDFS; LinkedIn: 120 Billion relationships; 82 Hadoop jobs daily (IIRC); 16 TB of intermedia data.
  • Who knew DevOps could be so funny? Adam Jacob, CTO of Opscode, gave a hilarious talk at the Velocity conference on the true nature of DevOps. Warning: your neck may get sore from nodding in agreement so much and your belly may ache from laughing so much.
  • Click to read more ...


Paper: GraphLab: A New Framework For Parallel Machine Learning

In the never ending quest to figure out how to do something useful with never ending streams of data, GraphLab: A New Framework For Parallel Machine Learning wants to go beyond low-level programming, MapReduce, and dataflow languages with a new parallel framework for ML (machine learning) which exploits the sparse structure and common computational patterns of ML algorithms. GraphLab enables ML experts to easily design and implement efficient scalable parallel algorithms by composing problem specific computation, data-dependencies, and scheduling.  Our main contributions include: 

  • A graph-based data model which simultaneously represents data and computational dependencies. 
  • A set of concurrent access models which provide a range of sequential-consistency guarantees. 
  • A sophisticated modular scheduling mechanism. 
  • An aggregation framework to manage global state. 

Click to read more ...