Navneet Kumar's Library tagged Architecture → View Popular

Filter:
All
Annotated

21 Dec 09

Java Virtual Machine's Internal Architecture

www.artima.com/...jvmP.html - Preview

Java JVM Virtual-Machine Architecture

08 Dec 09

Presentation Summary “High Performance at Massive Scale: Lessons Learned at Facebook” « Idle Process

idleprocess.wordpress.com/...le-lessons-learned-at-facebook - Preview

Performance Scalability FaceBook High-Performance Web2.0 Architecture Design

each user access requires retrieving data corresponding to user state
spread across hundreds of machines. Intra-cluster network performance
is hence critical to site performance
To deal with the significant slow down that resulted by synchronized loss in
relatively small TCP windows, Facebook built a custom congestion-aware UDP-based
transport that managed congestion across multiple requests rather than within a
single connection
8 more annotations...
- there are no joins in the code
- Storing dynamically changing data in a central database should be avoided.
- heavily-referenced static data should not be stored in a central database
- Logical migration of data is very difficult
- Data driven schemas make for happy programmers and difficult operations
- However, without synchronous updates, consecutive requests to the
  same data item from the same user can return inconsistent or stale results.
- he approach taken at Facebook is to set a cookie on user update requests
  that will redirect all subsequent requests from that user to the West coast
  master for some configurable time period to ensure that read operations do
  not return inconsistent results
- ”Work fast and don’t be afraid to break things.” Overall, the idea
  to avoid working cautiously the entire year, delivering rock-solid
  code, but not much of it

30 Sep 09

Microsoft SQL Server Development Customer Advisory Team : Deploying SQL Server 2005 with SAN #1

http://blogs.msdn.com/sqlcat/archive/2005/10/11/479887.aspx
http://blogs.msdn.com/sqlcat/archive/2005/11/21/495440.aspx
http://blogs.msdn.com/sqlcat/archive/2005/11/17/493944.aspx

blogs.msdn.com/...479887.aspx - Preview

Storage SAN SQL-Server Deployment LUN Infrastructure Architecture Configuration MSDN RAID

number of physical disks backing each LUN and which LUNs share the same physical
spindles.
Problems can arise when
different servers share the same physical spindles and have very different IO
characteristics (i.e. Exchange and SQL Server)
1 more annotations...
- Our recommendation is to use RAID 10 both for logs and
  data/index files. When cost is an issue you could allocate data and index files
  on RAID 5 or equivalent; however, keep logs on RAID 10.

Microsoft SQL Server Development Customer Advisory Team : SQL Server 2005 Configuration Blog #2.doc

blogs.msdn.com/...495440.aspx - Preview

Storage SAN SQL-Server LUN Infrastructure Architecture MSDN RAID

We recommend that you allocate a database log file to
a LUN which has dedicated physical disks that are not shared by other LUNs
including those which will be used for other log files or data / index objects.
The primary reason is that the log I/Os are sequential in nature and any other
disk activity can increase the log write latency
a set of dedicated spindles are used to create LUNs that are assigned to logs.
1 more annotations...
- If the work load has significant TempDB write activity (creation of temporary
  objects, row versioning, on-line index, etc.), it is preferred to have RAID 10
  for TempDB.

29 Sep 09

Facebook | Needle in a haystack: efficient storage of billions of photos

READ WRITE Performance

www.facebook.com/note.php - Preview

FaceBook Infrastructure Architecture Hardware Performance Read-Write

Since each image is stored in its own file, there is an enormous amount of
metadata generated on the storage tier due to the namespace directories and file
inodes. The amount of metadata far exceeds the caching abilities of the NFS
storage tier, resulting in multiple I/O operations per photo upload or read
request. The whole photo serving infrastructure is bottlenecked on the high
metadata overhead of the NFS storage tier, which is one of the reasons why
Facebook relies heavily on CDNs to serve photos
RAID-6 partition managed by the hardware RAID controller. RAID-6 provides
adequate redundancy and excellent read performance while keeping the storage
cost down
1 more annotations...
- The poor write performance is partially mitigated by the RAID controller NVRAM
  write-back cache. Since the reads are mostly random, the NVRAM cache is fully
  reserved for writes. The disk caches are disabled in order to guarantee data
  consistency in the event of a crash or a power loss

27 Jul 09

How FriendFeed uses MySQL to store schema-less data - Bret Taylor's blog

a "schema-less" storage system on top of MySQL

bret.appspot.com/...how-friendfeed-uses-mysql - Preview

Database MySQL Key-Value-Store Document-Database Architecture

making schema changes or adding indexes to a database with more than 10 - 20
million rows completely locks the database for hours at a time. Removing old
indexes takes just as much time, and not removing them hurts performance because
the database will continue to read and write to those unused blocks on every
INSERT, pushing important blocks out of memory
InnoDB stores data rows physically in primary key order. The
AUTO_INCREMENT primary key ensures new entities are written
sequentially on disk after old entities, which helps for both read and write
locality (new entities tend to be read more frequently than old entities since
FriendFeed pages are ordered reverse-chronologically