Friday, October 7, 2011

Astrolabe Cultists Nukes Science and History


Astrolabe Cultists Nukes Science and History

No good deed goes unpunished. What is a "labe", anyway? Who is the Astro?

How Network Management Systems Keep Time

At the core of every network management system is a critical component: time. ICMP and SNMP Polls are sent out according to a standard time interval. SNMP Traps are received on time intervals. Threshold crossings occur on certain time intervals. Reporting of these results, across timezones spanning a globe is quite complex.

If you have an alert in India, how do you know what the offset is from the reporting location? How does a web browser know the time offset from the web server location? When did the Daylight Savings Time change occur under President G.W. Bush across the various timezones in the United States? When a nation or city changes it's name - how does one know what time it is [before and after the name change] in those locations, in relation to the rest of the world?

Well, all of these political and geographic adjustments are kept in a timezone database.

A Short History of Time

It may be hard to wrap one's brain around why astrological cultists sued a coder who wrote software and key-punch operator who inserted information compiled from worldwide sources into a community database, free of charge, for decades of their life.

The timezone database dates back to at least November 24, 1986. As political changes occur, new versions of the database were published several times a year.

About 10 years later, Astrolabe advised Astro Computing Services on a Windows 3.1 PC user interface related to astrology, latitude, longitude, and timezones. (Microsoft Window 3.1 was released during March of 1992.) Astrolabe bought distribution rights to the ACS Windows 3.1 PC Atlas in 2008, which contains astrology information.

The Internet Engineering Task Force started the transition of the timezone database to international standards bodies on January 27, 2011. Arthur David Olston planned to retire from his unpaid position in managing the timezone database according to a post on slashdot on March 6, 2011.

Cultists at Astrolabe sent a letter to the retiring Arthur David Olston to take down the database from a U.S. government web site around May of 2011 (ironically, Astrolabe is unsure of the exact date, as per their court filing.) The cultist trolls at Astrolabe then filed a personal law suit against the retiring timezone keeper and the person who inserts new data into the database on September 30, 2011.

The mailing list was shut down on October 6, 2011. The FTP site, which contained information compiled by the community who participated in the mailing list was also shut down. Global systems with automated FTP accesses are now broken.

The open community reference for worldwide timezone information has been nuked. Clearly, this group who bought rights to a Windows 3.1 astrology program is little more than a useless copyright troll.

What Was Nuked: OUR Timezone History

Arthur David Olson, who wrote the code to parse the timezone database, has not earned a dime from his years of ongoing research and creation of the systematic timezone database that nearly all systems use world-wide - yet the cultists at Astrolabe are filing law suits against him, personally???

Paul Eggert, the contributor who made the "uniform naming convention" of [region]/[city] is also named personally in the law suit. He is the person who makes new changes to the database, as they are discovered world-wide from various political changes. How odd is suing an individual, who is making additions to a database???

To give you an understanding as to the scholarship of these individuals, one must read this from Jon Udell. There were many contributors to this database - the notes truly compose an interesting literary work, which goes back to the explanation of how and why time is accounted for today. For example, the blog included this particular quote from the nuked time database:

From Paul Eggert (2001-05-30):

Howse writes that Alaska switched from the Julian to the Gregorian calendar, and from east-of-GMT to west-of-GMT days, when the US bought it from Russia. This was on 1867-10-18, a Friday; the previous day was 1867-10-06 Julian, also a Friday. Include only the time zone part of this transition, ignoring the switch from Julian to Gregorian, since we can’t represent the Julian calendar.

As far as we know, none of the exact locations mentioned below were permanently inhabited in 1867 by anyone using either calendar. (Yakutat was colonized by the Russians in 1799, but the settlement was destroyed in 1805 by a Yakutat-kon war party.) However, there were nearby inhabitants in some cases and for our purposes perhaps it’s best to simply use the official transition.

Erasing decades of history and shooting old people and keypunch operators, who are also scholars in their own right, are not ways for cultists to win converts.

Call to Inquire of Purchase

If you need a timezone database, you should inquire if they have a substitute that adheres to the global TZ database standard for your software or operating system. If they don't offer a product for your operating system or adhering to the standard format, then tell them to get with the program.
Eastern Sales/Information: 1 (800) 843-6682

One can't definitively determine what "Eastern" time is, as labeled in their contact information page, since they nuked the timezone database. But one could always ask how they expect people to know what timezone they are in when their lawsuit was personally responsible for the removal of the timezone database from the internet.

Tech Support/Business Line: 1 (508) 896-5081
Hours 11am - 4pm
Hours 9:30am - 5:30pm
Eastern Fax: 1 (508) 896-5289
Snail Mail: Astrolabe Inc.
PO Box 1750
Brewster, MA 02631 USA

Some emails of interest.

Director of Marketing: Madalyn Hillis-Dineen
Technical Support: Support@alabe.com
Order questions: orders@alabe.com
Sales and nonTechnical questions: astrolabe@alabe.com
Webmaster: webmaster@alabe.com

Wednesday, October 5, 2011

Solaris 11 Preview (Part 2)


Solaris 11 Preview (Part 2)

Timothy Prickett Morgan from The Register wrote an excellent article "Oracle previews Solaris 11, due in November" which spurred some contemplations.

Solaris 11 Performance Features


In the "Pushing Solaris to 11" section, TPM talks about the features added to Solaris 11.

One of the things that probably took some extra time was doing optimizations in Solaris specifically for Oracle software, which Fowler touched on here

I honestly don't think that these features pushed out Solaris 11 release. Many of them appeared in Solaris 10 Update 10, so I don't think the Solaris 11 "extra time" was necessarily due to many of these features.

Solaris 10 Update 10 Performance Features

Some of the features mentioned in the "New in Oracle Solaris 11" happened to make it into Solaris 10 Update 10 - which many in the market are very pleased about.

In particular, some of the items included: ISM, DISM, RDSv3, many improvements in various threading libraries.

Other features, not mentioned in the release notes, but mentioned in Solaris 10 Update 10 patch notes include: T4 support, low latency socket framework


[time slider illustration]
Solaris 11 Upgrade Features

TPM also mentioned Fowler speaking about upgrade ease.
Solaris 11 will, for instance, have one button rollback features so if customers don't like the results of their upgrade to Solaris 11
With Solaris 11's integration with ZFS, multiple boot environments consume virtually no disk space, and upgrades can be done where the old boot environments may be allowed to remain in place (indefinitely, without losing OS mirroring capability), as with old Live Upgrade requirements. It just takes a keystroke to select the old boot environment under Solaris 10 Update 10 or Solaris 11.

There is an additional "time slider", which is part of OpenSolaris, Solaris 11 Express, Solaris 11, and various other Solaris derivative operating systems - being able to turn back time on a system is pretty wonderful. Fowler may have been alluding to this.

Solaris 10 Update 10 Upgrade Features

One of the features mentioned in the Solaris 10 Update 10 was the Live Upgrade of Solaris 10 using ZFS. I classified this as a "performance" enhancement, since Live Upgrade existed under Solaris 10 with easy rollback in the past, but it required a mirrored disk.

The Live Upgrade enhancement using ZFS is now here, for Solaris 11. The time-slider, from Solaris 11 is not.

Solaris 11 Preview (Part 1)


Solaris 11 Preview (Part 1)

Timothy Prickett Morgan from The Register wrote an excellent article "Oracle previews Solaris 11, due in November" which spurred some contemplations.
Fowler said that the current Solaris 10 tops out at 512 threads and a few terabytes of addressable main memory.
This may have been what Fowler said, but I wonder how much of an impacts of threads and large memory really has on Solaris 10, today. For example, the T3-4, which was released some time back, already surpasses 512 threads, with Solaris 10 performing linearly.
It is therefore not a coincidence that last year's top-end Sparc T3-4 server, which had eight 16-core, 128-thread Sparc T3 processors topped out at four sockets.
This years top-end SPARC T4-4 server halved the cores, halved the threads, maintained throughput, so the T4 processor could have been extended to 8 sockets with additional threads being added over the T3-4 platform. This being said, I think it was a coincidence. I don't really think the number of sockets is a Solaris issue. The math just does not suggest this conclusion.

I suspect the decision to top-out at 4 sockets has more to do with chip design realities. The decision to extend the cross-bar to manage another bit from 8 cores to 16 cores was a massive undertaking in the T3. The decision to extend add the S3 core into the T4 was a massive undertaking. The decision to integrate the massive S3 infrastructure onto the massive T3 crossbar will require more space, requiring a process shrink, which will also add significant effort.

Why does a process shrink require significant effort, when you are just making everything smaller? The answer to this question is related to another question which is always asked, "now that this is smaller, how much can we bump up the clock rate"?

The answer to this question deals with the length of the longest wires in the chip after the shrink - they are significant clock rate limiting factors. Any correction requires re-routing the "long" wires. The re-routing, by massive computing systems, absorb immense quantities of time. The chip testing, when adding functionality, also requires immense quantities of time.

These quantities of time are limiting factors. A single bug or manufacturing defect may require the turning-off of a feature (to be emulated in software, a-la Intel Floating Point bug) or a re-design of a chip area, without impacting surrounding areas.

Solaris 10 and 2011 Roadmaps



The early 2011 roadmap indicates Solaris 10 will experience a release in 2011 as well as 2012 - and the market just experienced Solaris 10 Update 10 release, according to the timeline provided.


The late 2011 roadmap did not show the on-time 2011 Solaris 10 release or the projected 2012 release... but there is an indication of the next T processor being pulled into the same timeframe as the 2012 Solaris 10 release.

Doing the math, it seems the "T5-8" platform (if Oracle remains consistent with their naming conventions) may really increase the number of simultaneous threads, and one might suspect that the 2012 projected "Solaris 10 Update 11" release might probably support the larger number of total number of threads.
2010 T3-4: 16 cores * 8 threads * 4 sockets = 1024 threads
2011 T4-4: 08 cores * 8 threads * 4 sockets = 0512 threads
2012 T5-8: 16 cores * 8 threads * 8 sockets = 2048 threads
Conclusion:

The migration to Solaris 11 will probably take a long time. The market is pretty happy to see many Solaris 11 performance features being back-ported to Solaris 10, especially since Solaris 11 is not supported under older SPARC hardware. More on this in Part 2 of this series.

Wednesday, September 28, 2011

SPARC T4: Launched 2011-09-26

Link
[Oracle SPARC T4 Processor]

SPARC T4: Launched 2011-09-26


Product Launch

The Register covered the T4 product launch in an article as well as the Oracle's SPARC SuperCluster in an additional article. The writer was kind enough to include images of the new systems being released based upon the new SPARC T4 silicon.

A simple PDF data sheet on the new Oracle SPARC T4 processor is available on Oracle's web site. On Monday, September 26, 2010 - a presentation and product launch was conducted. The Event Replay is available on-line, but some of the highlights are as follows.


SPARC T4 Announcement

Some features outlined included:
  • high throughput: 8 cores, 8 threads per core
  • high thread performance: clock rate: 2.85GHz and 3.00 GHz.
  • high-thread performance: single thread can use whole core
  • high thread performance: Out of Order execution
  • 4x on-chip DDR3 Memory Controller Channels
  • 2x on-chip 10GigE networking
  • 2x on-chip x8 PCIe 2nd generation I/O interfaces
  • 18 on-chip crypto engines


Price/Performance Comparison

Some price/performance comparisons were made between a T4-4 cluster and IBM POWER 7 system.



2010 Roadmap Reminder

The market was reminded of the 2010 SPARC processor road map and how other vendors refused to release a public road map.

IBM recently released a historical road map with no dates for POWER, but it seem they might be behind.



2011 Roadmap Prediction

Oracle illustrated how SPARC is beating their road map:
  • New T4 processor in advanced customer installations today
  • New T4 processor shipping today for normal customers
  • New T5? processor for 2012, delivery projected 6-12 months early
  • New T6? processor for 2013


2012 SPARC T5?

Details for the next processor, SPARC T5? scheduled to arrive next year in 2012, revealed.

On-Chip Enhancements
  • More crypto enhancements
  • Oracle RDBMS "numbers" acceleration
  • Hardware Decompression (I requested this a few years back!)
  • Memory Versioning (is this Transactional Memory?)
  • Low Latency Clustering
Performance Enhancements
  • Higher core clock rate
  • Multiple pipelines per core
  • More Cores per Socket
  • Larger chip caches
  • More memory bandwidth

Friday, September 23, 2011

Revisited: Oracle Database Licensing

Abstract:
Oracle licenses it's RDBMS by several factors, typically the Standard License (by socket) and an Enterprise License (by core scaling factor.) Occasionally, hardware and operating system vendors will enhance their offerings, requiring a revisit by database vendors to expand their legal categorizations for licensing. Oracle's guiding documents are readily available on-line.

Reason for Revisit:
A fairly extensive set of documents were posted over the past few years, but the URL's to these documents have changed, roughly since the acquisition of Sun by Oracle. Those core documents are listed below with why one might need to reference them.

Software Investment Guide
The Oracle Software Investment Guide is perhaps the most thorough document on performing Oracle installations within an organization. Perhaps these few sentences from the guide best describes what it contains.
We provide a detailed overview on how to license all Oracle products, from the Oracle database platform and application server to all Oracle enterprise applications, which includes Oracle E-Business Suite, Siebel, PeoplSoft, and JD Edwards.

Nine easy-to-read sections enable you to find the topics that interest you most. Within every section, each specific topic has hyperlinks and toll free numbers that enable you to get more information on the subject matter.

Additionally, we've included illustrations to enhance your understanding of our pricing practices related to data environments, batch processing, and more.

Pricing information on Oracle services, such as Support, Outsourcing, Consulting, and Education is also here.

This guide is clearly not isolated to merely Oracle Databases, but it is an authoritative source.

Database Licensing:
Oracle regularly updates their Database Licensing guideline document. Terminology such as Test, Production, Sockets, and Processors are defined within this document. Finer details which are often commonly asked questions include: development databases are normally no charge, but test and production databases must be purchased; Oracle Standard Editions may be charged by socket, but more advanced features means per-processor or per-user licensing must occur with Oracle Enterprise Edition; when dealing with IBM & Intel CPU's, "each chip in the multi-chip module is counted as one occupied socket."

Partitioning:
Oracle regularly updates their Server/Hardware Partitioning document, as system vendors create new technologies. Rigorous vendors create technologies categorized under "Hard Partitioning" (i.e. Capped Solaris Containers) while less rigorous vendors often create "Soft Partitioning" technologies (i.e. VMWare.) Some virtualization technology can be implemented as both Hard or Soft Partitioning, so implementation details must be attended to (i.e. Oracle VM implementation notes.) Costs can be controlled through careful architecture decisions, if one understands how a "Processor" is counted in such virtual environments.

Processor Core Factors:
Oracle regularly updates their Processor Core Factor Table, as new CPU designers release new central processor units. Certain multi-core CPU sockets with close throughput but vastly different core counts often have very different pricing (i.e. 16 core SPARC T3 "0.25" vs 8 core SPARC T4's "0.50" factor.)

Commercial Price List:
Oracle's standard Commercial Technology Price List which includes the database packages, is also published in PDF. While this price list changes regularly, Oracle specifies the retail pricing for databases such as: Standard Edition One, Standard Edition, Enterprise Edition, additional add-on components, etc.

In Conclusion:
It would be wise to track these changes to these documents, as new purchases are required, and new architectures are being developed.

Sunday, September 18, 2011

Solaris 10: Update 10 - Released!


Solaris 10: Update 10 - Released

Solaris 10 Update 10 has been released, note the "What's New" document.

Performance Enhancements Include:
  • Faster ZFS performance
  • Faster boot, with option to skip POST
  • Faster upgrades: Live Upgrade on ZFS root partition, instead of mirrors
  • Faster installation: use Flash Archive on ZFS root using text mode
  • Faster reboot: option to skip some POST tests
  • Enhanced throughput leveraging smt_pause for busy-wait or idle loops
  • Enhanced throughput: 64 bit libmtmalloc, atomic operations replace locks
  • Faster memory: 64 bit libmtmalloc, matrix lookups replace linked list
  • Use libmtmalloc for applications with threads>=16
  • Use libumem for applications with threads <16
  • Disk I/O improvement for SSD's with poor read-modify-write firmware
  • Improved I/O performance under Nahalem-EX platforms
  • Improved floating point, audio, video processing with Intel AVX Support
  • Faster Oracle startup/shutdown with ISM and DISM improvements
  • Faster high-speed networking for x86 bnx driver Jumbo Frame support
  • Increased network performance through bge driver MSI support
  • Faster RAC 11g via Remote Direct Memory Access interfaces via RDSv3
ZFS Improvements Include:
  • Set file system properties in a sent/received snapshot stream
  • ZFS Diff: Determine differences between snapshots
  • ZFS Recovery: Import a pool missing an intent log file
  • ZFS Recovery: Import a damaged pool read-only
  • ZFS Synchronous Behavior: standard, always, disabled
  • Interval and Count for zpool: list and status
New Capabilities Include:
  • Greater than 2 terabyte memory support
  • Solaris Volume Manager clustered dataset import for recovery
  • User extensibility for up to 1024 group membership
  • SaMBa upgrade to 3.5.8
  • x86 BASH upgrade to 3.2
  • Apache C++ standard library upgrade to 4
Observability Enhancement Include:
  • ZFS Difference capability
  • Processor Group (pginfo) OS performance hardware view
  • Processor Group (pgstat) hardware and software utilization
  • Logical disk name to JBOD or blade chassis bays (diskinfo) view
Security Enhancements Include:
  • LDAP name service tol functionality without LDAP being configured
  • IVv6 NAT on IPFilter
  • Solaris Cryptographic and Key Management Framework support for OKM
  • Additional AES cyphers in kernel SSL (Secure Sockets Layer)
  • New Passwords no longer unlock accounts
  • Password construction policy applies to root by default
  • SSH supports chroot capability
Device Support Enhancements Include:
  • Intel ixgbe driver support for: 82599, x540/x540T
  • Intel igb driver support for: 82576, 82580, i350
  • Intel e1000g 82579LM/LF LOM controllers: 1502,1506, 1519
  • Broadcom bge driver support: BCM5717, BCM5724
  • QLogic P3+ FCoE CNA Ethernet qlcnic and device support
  • Mellanox ConnectX-2 10GigE mcxnex/mcxe drivers for GLDv3
  • Intel Patsburg SCU support: ld60, ld61, ld64, ld65, ld68, ld69
  • LSI MegaRAID Falcon SAS 2.0 HBA device via imraid_sas
  • LSI SAS 2308 HBA storage device via mpt_sas
  • LSI SAS 2208 HBA device via mr_sas
  • Public Generic LAN Driver interface for bge

Wednesday, September 14, 2011

Oracle License Change: Add SPARC T4


(Oracle SPARC T4 micrograph)

Oracle License Change: Add SPARC T4

Abstract

Oracle licenses it's RDBMS by several factors, typically the Standard License (by socket) and an Enterprise License (by core scaling factor.) Occasionally, Oracle will change the core scaling factor, resulting in discounting or liability for the consumer.

The Platform

The SPARC CPU from Oracle is an implementation of the SPARC V9 open specification. There have been several series of chips based upon this implementation: T1, T2, T2+, and T3. The T1 & T2 are both single socket implementations, while the T2+ and T3 are a multi-socket implementation. Oracle has released on their roadmap that the SPARC T4 processor will be coming out shortly.

The Addition

The SPARC T4 has been added to the Oracle RDBMS "Processor-Core Factor Table".

Factor Vendor/Processor
0.25 SUN UltraSPARC T1 <1.4GHz
0.25 Oracle SPARC T3
0.50 SUN UltraSPARC T1 1.4GHz
0.50 SUN UltraSPARC T2+ Multicore
0.50 Fujitsu SPARC VII+
0.50 Oracle SPARC T4
0.75 SUN UltraSPARC IV, IV+, or earlier
0.75 Fujitsu SPARC64 VI, VII
0.75 SUN UltraSPARC T2


Note, Green is new. Oracle has added the T4 processor with a core factor of 0.50.

Impacts to Network Management Infrastructure

It appears that Oracle will be releasing SPARC T4 into production. Purchasing should be watched very closely during this transition period.

If there is the need for increased thread performance, the 8 core T4 may be leveraged instead of the 16 core T3, with no impact to Oracle licensing when databases require an RDBMS.

The UltraSPARC IV+ 21.GHz processors had some of the best single-threaded performance characteristics of any SPARC processor ever produced (albeit, the throughput pales in comparison to any of the newer generation of multicore SPARC systems.)

The Oracle SPARC T4 finally looks like a good candidate to replace those old SUN UltraSPARC IV+ systems, which are so highly cherished for their single thread performance. There is probably no better Network Managment platform to consider at this point in time.