Overturn of HFT Market – (High frequency trading)

July 29, 2013 2 comments

High-frequency trading (HFT) is the use of sophisticated technological tools and computer algorithms to transact a large number of orders at very fast speeds. HFT uses complex algorithms to analyze multiple markets and execute orders in seconds or fractions of a second  based on market conditions. High-frequency traders usually do not hold their portfolios overnight, they move in and out of short-term positions aiming to capture sometimes just a fraction of a cent in profit on every trade, they typically compete against other HFTs, rather than long-term investors. HFT is accounted for more than 50 percent of all U.S. equity volume.

The trading market is very dynamic and competitive, by 2007 Unix was the market standard, in 2008 started the adoption of Linux in trading market, due the Linux kernel’s ability to pass messages very quickly, then it became the standard between 2008 and 2012. In early of 2013 Oracle announced  its new line of Sparc servers  with T5 processor and it has been changing the market again.

In order to offer a competitive HFT solution to the market, it is important to care about the hardware where the solution will be running, the software should be developed to use the all advanced hardware and operational system benefits. Oracle’s processor offers resources like software on silicon, where some algorithms are put out of the software and placed into hardware besides the operational system Solaris that takes full advantage of the hardware features and can provide real time features to the trade’s process.

Another important point is about the data, the trade process should be able to handle data in real-time, therefore a solution to cache the data in memory is crucial to a HFT process.

Oracle Coherence is the industry leading in-memory data grid solution that enables organizations to predictably scale mission-critical applications by providing fast access to frequently used data.

The main reason of the market changing is the integration of hardware and software that only Oracle can offer.

Categories: Java, Oracle

Software on Silicon

June 20, 2013 Leave a comment

Sparc T5 Chip

Oracle announced in early 2013 its new Sparc T5 processor, it is twice as fast as their T4 predecessors and the world’s fastest microprocessor for database and java application.

Oracle’s new SPARC T5 servers have set 17 world records and are the world’s best platforms for enterprise computing at any scale, delivering the best value for database and enterprise applications.

Sparc T5 has 16 cores, each core can support up to eight independent threads or strands, a Sparc Server scales up to 8 sockets, so for the operating system there are a total of 1024 CPUs.

According to Oracle Sparc processor roadmap, each generation will have 2x performance than its prior generation, they will do that due Software on cilicon concept, where the algorithm is put out of software, and placed into hardware, they have done this for encryption/decryption on T4/T5 processors, and for the next generation they will do it for:

  • Database Query
  • Compression
  • Encryption
  • Cluster Interconnect
  • Application Data Protection

The following graph represents the comparison of end-to-end overall application throughput performance including both SSL and WS-Security operations using hardware-assisted (Software on silicon) cryptographic acceleration and without using them. The tests ran a Java EE/JAX-WS Web Services application using a 500k XML message payload:

Figure 1

The results showed using Oracle SPARC hardware-assisted cryptographic acceleration significantly contributed to combined SSL/TLS and WS-Security scenarios delivering more than 200% of overall application performance gain for Two-way SSL and WS-Security operations in comparison with SSL/WS-Security operations using software.

Categories: Oracle

Scripts to monitoring the Oracle ASM disks, files and volumes

October 24, 2012 6 comments

This is a set of scripts to help on Oracle ASM Monitoring

Space used in ASM Disk Groups

SET LINESIZE  145
SET PAGESIZE  9999
SET VERIFY    off
COLUMN group_name             FORMAT a20           HEAD 'Disk Group|Name'
COLUMN sector_size            FORMAT 99,999        HEAD 'Sector|Size'
COLUMN block_size             FORMAT 99,999        HEAD 'Block|Size'
COLUMN allocation_unit_size   FORMAT 999,999,999   HEAD 'Allocation|Unit Size'
COLUMN state                  FORMAT a11           HEAD 'State'
COLUMN type                   FORMAT a6            HEAD 'Type'
COLUMN total_mb               FORMAT 999,999,999   HEAD 'Total Size (MB)'
COLUMN used_mb                FORMAT 999,999,999   HEAD 'Used Size (MB)'
COLUMN pct_used               FORMAT 999.99        HEAD 'Pct. Used'

break on report on disk_group_name skip 1
compute sum label "Grand Total: " of total_mb used_mb on report

SELECT
    name                                     group_name
  , sector_size                              sector_size
  , block_size                               block_size
  , allocation_unit_size                     allocation_unit_size
  , state                                    state
  , type                                     type
  , total_mb                                 total_mb
  , (total_mb - free_mb)                     used_mb
  , ROUND((1- (free_mb / total_mb))*100, 2)  pct_used
FROM
    v$asm_diskgroup
ORDER BY
    name
/

Disks contained within Disk Groups

SET ECHO        OFF
SET FEEDBACK    6
SET HEADING     ON
SET LINESIZE    180
SET PAGESIZE    50000
SET TERMOUT     ON
SET TIMING      OFF
SET TRIMOUT     ON
SET TRIMSPOOL   ON
SET VERIFY      OFF

CLEAR COLUMNS
CLEAR BREAKS
CLEAR COMPUTES

COLUMN disk_group_name        FORMAT a25           HEAD 'Disk Group Name'
COLUMN disk_file_path         FORMAT a20           HEAD 'Path'
COLUMN disk_file_name         FORMAT a20           HEAD 'File Name'
COLUMN disk_file_fail_group   FORMAT a20           HEAD 'Fail Group'
COLUMN total_mb               FORMAT 999,999,999   HEAD 'File Size (MB)'
COLUMN used_mb                FORMAT 999,999,999   HEAD 'Used Size (MB)'
COLUMN pct_used               FORMAT 999.99        HEAD 'Pct. Used'

BREAK ON report ON disk_group_name SKIP 1

COMPUTE sum LABEL ""              OF total_mb used_mb ON disk_group_name
COMPUTE sum LABEL "Grand Total: " OF total_mb used_mb ON report

SELECT
    NVL(a.name, '[CANDIDATE]')                       disk_group_name
  , b.path                                           disk_file_path
  , b.name                                           disk_file_name
  , b.failgroup                                      disk_file_fail_group
  , b.total_mb                                       total_mb
  , (b.total_mb - b.free_mb)                         used_mb
  , ROUND((1- (b.free_mb / b.total_mb))*100, 2)      pct_used
FROM
    v$asm_diskgroup a RIGHT OUTER JOIN v$asm_disk b USING (group_number)
ORDER BY
    a.name
/

List of ASM files as well as  its volume information

SET ECHO        OFF
SET FEEDBACK    6
SET HEADING     ON
SET LINESIZE    180
SET PAGESIZE    50000
SET TERMOUT     ON
SET TIMING      OFF
SET TRIMOUT     ON
SET TRIMSPOOL   ON
SET VERIFY      OFF

CLEAR COLUMNS
CLEAR BREAKS
CLEAR COMPUTES

COLUMN full_path              FORMAT a75                  HEAD 'ASM File Name / Volume Name / Device Name'
COLUMN system_created         FORMAT a8                   HEAD 'System|Created?'
COLUMN bytes                  FORMAT 9,999,999,999,999    HEAD 'Bytes'
COLUMN space                  FORMAT 9,999,999,999,999    HEAD 'Space'
COLUMN type                   FORMAT a18                  HEAD 'File Type'
COLUMN redundancy             FORMAT a12                  HEAD 'Redundancy'
COLUMN striped                FORMAT a8                   HEAD 'Striped'
COLUMN creation_date          FORMAT a20                  HEAD 'Creation Date'
COLUMN disk_group_name        noprint

BREAK ON report ON disk_group_name SKIP 1

COMPUTE sum LABEL ""              OF bytes space ON disk_group_name
COMPUTE sum LABEL "Grand Total: " OF bytes space ON report

SELECT
    CONCAT('+' || db_files.disk_group_name, SYS_CONNECT_BY_PATH(db_files.alias_name, '/')) full_path
  , db_files.bytes
  , db_files.space
  , NVL(LPAD(db_files.type, 18), '<DIRECTORY>')  type
  , db_files.creation_date
  , db_files.disk_group_name
  , LPAD(db_files.system_created, 4) system_created
FROM
    ( SELECT
          g.name               disk_group_name
        , a.parent_index       pindex
        , a.name               alias_name
        , a.reference_index    rindex
        , a.system_created     system_created
        , f.bytes              bytes
        , f.space              space
        , f.type               type
        , TO_CHAR(f.creation_date, 'DD-MON-YYYY HH24:MI:SS')  creation_date
      FROM
          v$asm_file f RIGHT OUTER JOIN v$asm_alias     a USING (group_number, file_number)
                                   JOIN v$asm_diskgroup g USING (group_number)
    ) db_files
WHERE db_files.type IS NOT NULL
START WITH (MOD(db_files.pindex, POWER(2, 24))) = 0
    CONNECT BY PRIOR db_files.rindex = db_files.pindex
UNION
SELECT
    '+' || volume_files.disk_group_name ||  ' [' || volume_files.volume_name || '] ' ||  volume_files.volume_device full_path
  , volume_files.bytes
  , volume_files.space
  , NVL(LPAD(volume_files.type, 18), '<DIRECTORY>')  type
  , volume_files.creation_date
  , volume_files.disk_group_name
  , null
FROM
    ( SELECT
          g.name               disk_group_name
        , v.volume_name        volume_name
        , v.volume_device       volume_device
        , f.bytes              bytes
        , f.space              space
        , f.type               type
        , TO_CHAR(f.creation_date, 'DD-MON-YYYY HH24:MI:SS')  creation_date
      FROM
          v$asm_file f RIGHT OUTER JOIN v$asm_volume    v USING (group_number, file_number)
                                   JOIN v$asm_diskgroup g USING (group_number)
    ) volume_files
WHERE volume_files.type IS NOT NULL
/

Disks performance metric

SET ECHO        OFF
SET FEEDBACK    6
SET HEADING     ON
SET LINESIZE    256
SET PAGESIZE    50000
SET TERMOUT     ON
SET TIMING      OFF
SET TRIMOUT     ON
SET TRIMSPOOL   ON
SET VERIFY      OFF

CLEAR COLUMNS
CLEAR BREAKS
CLEAR COMPUTES

COLUMN disk_group_name    FORMAT a20                    HEAD 'Disk Group Name'
COLUMN disk_path          FORMAT a20                    HEAD 'Disk Path'
COLUMN reads              FORMAT 999,999,999,999        HEAD 'Reads'
COLUMN writes             FORMAT 999,999,999,999        HEAD 'Writes'
COLUMN read_errs          FORMAT 999,999,999            HEAD 'Read|Errors'
COLUMN write_errs         FORMAT 999,999,999            HEAD 'Write|Errors'
COLUMN read_time          FORMAT 999,999,999,999        HEAD 'Read|Time'
COLUMN write_time         FORMAT 999,999,999,999        HEAD 'Write|Time'
COLUMN bytes_read         FORMAT 999,999,999,999,999    HEAD 'Bytes|Read'
COLUMN bytes_written      FORMAT 999,999,999,999,999    HEAD 'Bytes|Written'

BREAK ON report ON disk_group_name SKIP 2

COMPUTE sum LABEL ""              OF reads writes read_errs write_errs read_time write_time bytes_read bytes_written ON disk_group_name
COMPUTE sum LABEL "Grand Total: " OF reads writes read_errs write_errs read_time write_time bytes_read bytes_written ON report

SELECT
    a.name                disk_group_name
  , b.path                disk_path
  , b.reads               reads
  , b.writes              writes
  , b.read_errs           read_errs 
  , b.write_errs          write_errs
  , b.read_time           read_time
  , b.write_time          write_time
  , b.bytes_read          bytes_read
  , b.bytes_written       bytes_written
FROM
    v$asm_diskgroup a JOIN v$asm_disk b USING (group_number)
ORDER BY
    a.name
/
Categories: Oracle

ASMLib for RHEL6

October 16, 2012 Leave a comment

Are you worried because you can’t find asmlib for RHEL 6?

don’t worry, this article clarifies that.

ASMLib

ASMLib is an optional set of tools and a kernel driver that can be inserted between ASM and the hardware, as well as an application library used by the Oracle database software to access ASM disks. It is a support library for the ASM feature of Oracle 10g and 11g single instance database servers as well as RAC installations. ASM and regular database instances can use ASMLib as an alternative interface for disk access. ASMLib has three components:

  • Kernel driver – oracleasm is a Linux kernel driver also known as the Oracle ASMLib kernel driver.
    This is an open-source (GPL) kernel driver and is available from Oracle as source and binary
    RPMs. Note that although this driver is provided under an open source license, it has not been
    accepted into the mainline Linux kernel.
  • Support tools – oracleasm-support provides the utilities to manage the ASM library driver. Oracleasm-support is an open-source package (GPL) and is available from Oracle as source and binary RPMs.
  • Application library – oracleasmlib package provides the actual ASM library. This is a closed source, binary-only RPM, available as a free download from Oracle.

Oracle introduced ASMLib in 2004 to provide Oracle workloads on Linux with performance and stability
comparable to equivalent workloads on UNIX. ASMLib addressed deficiencies – including a lack of async
and direct I/O — that existed in the 2.4 Linux kernel.

Kernel 2.6

The Linux community addressed these deficiencies in the 2.6 Linux kernel with the addition of udev and device-mapper multipath which together with LVM provided native multipathing, scalable native volume management, and persistent device naming. Additionally, the 2.6 Linux kernel added tools for handling large numbers of disks, especially SAN-attached disks.

Since the RHEL6’s kernel version is 2.6, the use of asmlib is no longer needed, in RHEL 6 is used UDEV and DirectIO instead.

UDEV

UDEV allows for rules that specify what device name is given to a specific device, regardless of which port is plugged into.

Direct I/O

When a process wants to access data from a file, the SO brings the data into a buffer cache in order to avoid frequency of disk accesses. Oracle database also keep the data in a area called Buffer Cache in SGA, this “doublecopying” of data results in more CPU consumption and adds overhead to the memory too, the use of Direct I/O  is indicated in situations like that, applications like a database that wishes to bypass the SO buffering within the file system cache.

To enable the use of DirectI/O on Oracle the parameter disk_asynch_io must be set to true, which is set to true by default in Oracle 11g:

SQL> show parameter disk_asynch_io;

NAME                                 TYPE        VALUE
------------------------------------ ----------- ------------------------------
disk_asynch_io                       boolean     TRUE
SQL>

Conclusion

The conclusion is that if you are using kernel 2.6 or higher the ASMLib is useless, there are native implementations(UDEV and DirectI/O) that replaced its needed, UDEV to persists the name of devices and the DirectIO to bypass the SO buffering.

Categories: Linux, Oracle

Tuning linux server for Oracle database

August 31, 2012 1 comment

Dozens of DBA and SysAdmin do not know how to calculate the kernel’s parameters correctly and most of times they use the default value as indicated in Oracle Documentation, I would say that most of the DBA do not even know the meaning of those parameters.

In this article I will describe the meaning of each OS paramater and how to calculate/sizing them.

Shared Memory

Shared memory allows processes to access common structures and data by placing them in shared memory segment. It is the fastest form of interprocess Communication (IPC) available since no kernel involvement occurs when data is passed between the processes. In fact, data does not need to be copied between the processes.

Oracle uses shared memory segments for Shared Global Area (SGA) which is an area of memory that is shared by Oracle processes. The size of the SGA has a significant impact to Oracle’s performance since it holds database buffer cache and much more.

Basically we use three parameters to scale shared memory:

SHMMAX

This parameter defines the maximum size in bytes of a single shared memory segment that a linux process can allocate.

SHMMNI

This parameter sets the system wide maximum number of shared memory segments. Oracle recommends SHMMNI to be at least 4096.

SHMALL

This parameter sets the total amount of shared memory pages that can be used system wide.

Memory Sizing

The steps bellow help to scale the parameters SHMMAX and SHMALL:

Obtain the total memory from the system

mem=$(free|grep Mem|awk '{print$2}')

Convert the value of $mem to bytes

totmem=$(echo "$mem*1024"|bc)

Get the Hugepagesize from /proc/meminfo

huge=$(grep Hugepagesize /proc/meminfo|awk '{print $2}')

Calculate what 75% of the total memory on the system for SHMMAX

max=$(echo "$totmem*75/100"|bc)

Divide the SHMMAX value by the Hugepagesize to get SHMALL

all=$(echo "$max/$huge"|bc)

Set the SHMMAX value in the /etc/sysctl.conf file

echo "kernel.shmmax = $max" >> /etc/sysctl.conf

Set the SHMALL value in the /etc/sysctl.conf file

echo "kernel.shmall = $all" >> /etc/sysctl.conf

Setting the maximum number of shared memory segments with SHMMNI.

echo "kernel.shmmni=4096" >> /etc/sysctl.conf

Memory Tuning

The linux kernel tunable parameter vm.swappiness (/proc/sys/vm/swappiness) can be used to define how aggressively memory pages are swapped to disk.

Linux moves memory pages that have not been accessed for some time to the swap space even if there is enough free memory available. By changing the percentage in /proc/sys/vm/swappiness you can control the swapping behavior, depending on the system configuration.

A high swappiness value means that the kernel will be more apt to unmap mapped pages. A low swappiness value means the opposite, the kernel will be less apt to unmap mapped pages. In other words, the higher the vm.swappiness value, the more the system will swap.

Swapping for Oracle is too bad

vm.swappiness=0

Maximum percentage of active memory that can have dirty pages

vm.dirty_background_ratio=3

Maximum percentage of total memory that can have dirty pages

vm.dirty_ratio=15

How long data can be in page cache before being expired (hundreths of a second)

vm.dirty_expire_centisecs=500

How often pdflush is activated to clean dirty pages (hundreths of asecond)

vm.dirty_writeback_centisecs=100

Setting Hugepages

Without HugePages, the memory of the SGA is divided into 4K pages, which have to be managed by the Linux kernel. Using HugePages, the page size can be increased to anything between 2MB and 256MB, thereby reducing the total number of pages to be managed by the kernel and therefore reducing the amount of memory required to hold the page table in memory. In addition to these changes, the memory associated with HugePages can not be swapped out, which forces the SGA to stay memory resident.

Setup the oracle user to be able to use hugepages in /etc/sysctl.conf

vm.hugetlb_shm_group=`id -g oracle`

Calculating the recommended number of HugePages, use the following formula:

(SGA+PGA+(20KB * # of Oracle processes running)) / 2MB

For example:

(20GB SGA + 10GB PGA + (20KB * 1,000 Oracle processes)) / 2MB = 15369

Include the parameter vm.nr_hugepages in /etc/sysctl.conf

vm.nr_hugepages=15369

Setting Semaphores

Semaphores are counters which are used to provide synchronization between processes for shared resources like shared memory. Oracle’s process concur in order to access the SGA, which is allocated in a shared memory area, thereby the number of semaphores must be higher than the number of oracle’s process parameter.

The number of semaphores per set can be defined through the kernel parameter SEMMSL. In /etc/sysctl.conf the entry for semphores is:

kernel.sem="250 32000 100 142"

The first value, SEMMSL, is the maximum number of semaphores per semaphore set

The second value, SEMMNS, defines the total number of semaphores for the system

The third value, SEMOPM, defines the maximum number of semaphore operations per semaphore call

The last value, SEMMNI, defines the number of entire semaphore sets for the system

Network tuning

Receive socket buffer size

net.core.rmem_default=262144
net.core.rmem_max=4194304

Send socket buffer size

net.core.wmem_default=262144
net.core.wmem_max=4194304

TCP socket buffer size

net.ipv4.tcp_rmem=4096 262144 4194304
net.ipv4.tcp_wmem=4096 262144 4194304

How often to send keep alive packets when a connection is unused

net.ipv4.tcp_keepalive_time=30

How long the kernel waits in between probes

net.ipv4.tcp_keepalive_intvl=60

How many probes are sent before a connection is considered broken

net.ipv4.tcp_keepalive_probes=9

How many times to retry before killing the connection

net.ipv4.tcp_retries2=3

How many times to retry transmitting the syn packet

net.ipv4.tcp_syn_retries=2
Categories: Linux, Oracle

Configuring Database Failover on JBoss

April 27, 2012 2 comments

JBoss supports database failover for datasources, this functionality is supported for both local datasources and XA datasources.

If the primary database becomes unavailable, it will try every database listed, until it finds one that sticks. If for a connection request it has tried all the URLs in the list, and none worked, an exception will be thrown with a message saying that none of the URLs worked. The current selection policy is round-robin and so will not revert to the primary database when it becomes available unless bounce the instance.

Basically what you must to do is configure some elements in datasource file, the following elements must be used to configure your failover datasource:

  • <ha-local-tx-datasource> - The root element for local datasources that support failover.
  • <url-delimeter>which contains a string that is used as a delimeter between URLs in connection-url.
  • <connection-url> - List of connection urls

Configure local datasource for failover:

   <ha-local-tx-datasource>
      <jndi-name>HADefaultDS</jndi-name>
      <connection-url>jdbc:oracle:thin:@server1:1521:jboss|jdbc:oracle:thin:@server2:1521:jboss|...</connection-url>
      <url-delimeter>|</url-delimeter>
      <driver-class>oracle.jdbc.OracleDriver</driver-class>
      <user-name>sa</user-name>
      <password></password>
   </ha-local-tx-datasource>

Configure XA datasource for failover:

   <ha-xa-datasource>
      <jndi-name>MockedHaXaDS</jndi-name>
      <track-connection-by-tx></track-connection-by-tx>
      <isSameRM-override-value>false</isSameRM-override-value>
      <xa-datasource-class>oracle.jdbc.xa.client.OracleXADataSource</xa-datasource-class>
      <xa-datasource-property name="URL">jdbc:oracle:thin:@server1:1521:jboss|jdbc:oracle:thin:@server2:1521:jboss|...</xa-datasource-property>
      <url-property>URL</url-property>
      <url-delimeter>|</url-delimeter>
   </ha-xa-datasource>
Categories: Middleware

How to hide password entry using stty

September 28, 2011 Leave a comment

This scripts allow the user to type the password without it be shown in the screen.

Copy and past the content bellow  in a .sh file

#!/bin/bash 

echo -n "Password: "
stty -echo
read PASSWD
stty echo
echo -e "\nYour password is:" $PASSWD

Running the shell script

# ./hiddenpasswd.sh
Password:
Your password is: t$prod123%f
Categories: Linux