Hello, my name is Manta

With this first post to our Remora web site, I would like to introduce myself and announce the appointment of Peter Holmes into Major Accounts.

Peter brings a commendable heritage of technology and solutions-selling expertise and experience, ranging from positions at Honeywell and DEC in the 80s to Fujitsu, Oracle, Borland and Technology One in more recent times. Peter will take on the exciting role of Major Accounts, continuing Remora's vision to bring the best technology to our ever-expanding client base.

As this is my first post, I should introduce myself. I am a director and co-founder of Remora, alongside Robert Silver, Chief Executive. I lead and manage the engineering team at Remora as Chief Technologist and partisan of engineering zeal. I like obscure English, occasionally verb nouns, am a stickler for quality control, sometimes to argument, and I say what I believe regardless of the consequences. I will make posts on this web site to announce and describe the Remora process and our offering, as well as occasionally whine and moan about what could be better in the world, most of the time about idiocracy in the world that holds us all back.

I'll be posting under my Remora alias Manta. All Remoreans have oceanic aliases for our cloud accounts and systems. This craziness extends to our servers and infrastructure, into the depths of our internal software, and even out into our customers' sites. Rob is Marlin, watch for posts from him, and Peter will probably pipe up at some point, he is Snapper.

As a result of Peter's appointment and a recent "coming of age" of Remora's capability and confidence, our engineering solution team and account management team will autonomously handle most Oracle and Splunk opportunities and customers, so I will now refocus on a new long term goal.

My new ambition is to bring data from physical plant and equipment to the forefront of business and operational decision-making. I believe that there is an incredible wealth of information going untapped - and I plan to apply the newest and most intelligent technology to this stream of "data exhaust" and enriching the data from non-computing equipment with traditionally disparate sources. I plan to radically institute the concept of "dimensionalising" machine and equipment data with (i) other machine data, (ii) garden-variety relational databases and computing-based data sources, and (iii) modern online sources of time-based data. The ultimate aim is to provide new levels of visibility, predictability and control of operational systems - enabling more efficient, more controlled, better understood and safer decision making.

Watch the tags on this site, they are dynamically weighted using a fancy algorithm in order to create a living home page that prioritises what Remora's team is working on or thinking about. Also please feel free to contact any of us about what we post or what you want to hear - it's fairly easy to figure out an email address from our aliases and our domain name (but we won't publish them for obvious reasons).

Production release of updated UEK

Oracle UEK 2.0 provides advanced features for production environments.

Notable inclusions in this release are as follows

The addition of btrfs (B-tree file system) with copy-on-write functionality which allows for improved performance in VM environments. Btrfs offers improvements in scalability, reliability and ease of management from the widely used ext4 filesystem. The primary benefits of btrfs allow for efficient cloning and snapshot capabilities including snapshot rollbacks.

Linux containers for OS level virtualisation is available on UEK2.0 enabling the running of multiple isolated Linux environments on a single host. Each container is a virtual environment with its own process and network space. Rather than having separate VMs each requiring OS administration, containers allow for a common host OS with isolated environments for services and applications.

DTrace for dynamic troubleshooting of kernel and application problems in real time has also been included. DTrace allows for the inspection of memory, CPU time, filesystem and network resources used system wide or by individual processes. Detailed inspection of individual processes is also possible with logging of arguments passed to functions or file access.

KSplice is incorporated which will allow for kernel updates of your production environments with zero downtime. KSplice allows you to keep up to date with the latest kernel security patches without the need to restart applications or reboot your server. Rollbacks are also possible without reboots should an update cause any unexpected issues.

The release of UEK 2.0 affirms Oracle's commitment to Linux mainline tree.

Unbreakable Enterprise Kernel Release 2 is included with Oracle Linux 5 and 6.

The latest addition to Oracle Engineered Systems

The Oracle Big Data Appliance joins Exadata, Exalogic, Exalytics, SPARC SuperCluster and the Oracle Database Appliance to further grow the scope of the Oracle Engineered Systems product family. Like the other systems in that family, the Big Data appliance combines software and hardware in a single offering to provide an out-of-the-box system engineered for extreme performance.

The appliance combines the new Oracle NoSQL database, Hadoop and Cloudera Manager to provide massive data storage, querying and processing capabilities. A connector for Oracle Database 11G also enables easy integration of this power into an existing environment. The frequent bottleneck in big data systems of moving a huge volume of data around is not forgotten with internal Infiniband connections between data nodes.

Full details are available in the press release:

• http://www.oracle.com/us/corporate/press/1453721?msgid=3-5784744152

Streamlining Linux configuration for Oracle deployments

Oracle has recently added a new rpm aimed at simplifying the deployment of RDBMS 10g or 11g on to Oracle Linux systems. This rpm--'oracle-validated'--carries out many tasks which will be familiar not only to Linux users, but also those who have successfully used Solaris JumpStart for much the same purpose.

In brief, it does the following:

• Installs the packages and dependencies needed to successfully run the database
• Creates the required users
• Modifies kernel parameters and other system settings as required

This automates a lot of the routine configuration needed for any Oracle database installation, and is definitely a win for hard-pressed sysadmins. You can find more details here:

• http://www.oracle.com/technetwork/articles/servers-storage-admin/ginnydbinstallonlinux-488779.html?msgid=3-5784744152

A few interesting new features

Oracle VM has just been updated to 3.03. The main new feature is anti-affinity support, which allows definition of which guests may not reside simultaneously on a single host. This is useful in cases such as a database cluster on guests, in which having multiple nodes on the same physical hardware would negate the hardware fault tolerance the cluster is providing.

Other notable improvements are mainly within Oracle VM Manager, which has had some extra polish added to the user interface.

A full changelog can be found at:

• http://docs.oracle.com/cd/E26996_01/e18546/index.html

Summary of information pertaining to Sun Blade technology and OVM

I have spent some time doing research into the use of Sun Blade 6000 series to run a OVM cluster, you can find links to various resources that I used at the bottom of this post.
But first some more details about Blade servers.

Blade servers are composed of several modules that are put into a chassis, in this case, the Sun Blade 6000 Chassis.
The chassis takes up 10RU of space, irrespective of the number of actual modules inserted into the chassis. There are currently 5 server modules, 2 SPARC and 3 x86.
There is no limitations to what modules you can mix and match, so in the same chassis can have all 5 server modules, with no issues.
There are also network express modules, these act as a kind of "switch", they are connected to all the server modules in the chassis, and provide additional network access ports of various flavours (depends on the type of network module).

Anyway, it would appear that using a blade set up would make developing a OVM solution a lot easier. The blades are cheaper that normal 1RU servers, they can be set up to be either servers or as storage, and provide a much reduced level of cabling and complexity as each blade can be set up to communicate with each other (by default, they run in "private mode" which prevents blades from listening to each others traffic).

A basic set up could be 2 X6270 server modules, each module can have a 2 processors, up to 144 GB of RAM and a couple of hard drives, along side a fully laden Vela2 Storage module (4.6TB of data storage).
Along with network connectivity components (hard to use a VM that no-one has access to), this set up would provide 2 VM machines, with the OVM manager as a VM, with a lot of RAM and processing power for VMs, as well as enough storage.

Overall, Blade servers are actually rather cheap, with the above set-up costing no more that $50,000 RRP, with the cost of additional servers being quite less. It is also possible to have OVM Server pre-installed on the X6270's, which would make setting up OVM overall quite easy, although then you would lack the fun setting it up yourself.

This isn't the end of my posts on the topic, just for now.

Links that may be helpful for further reading:
Sun Blade 6000 I/O and Management Architecture:

• http://www.oracle.com/technetwork/articles/systems-hardware-architecture/sb6000-io-mgmt-400403.pdf

Oracle Optimized Solution for Enterprise Cloud Infrastructure:

• http://www.oracle.com/us/solutions/opt-solutions-cloud-technical-wp-405951.pdf

Accelerating Deployment of Virtualized Infrastructures with the Oracle VM Blade Cluster Reference Configuration:

• http://www.oracle.com/us/technologies/virtualization/ovm-ref-bus-wp-192123.pdf

Best Practices and Guidelines for Deploying the Oracle VM Blade Cluster Reference Configuration:

• http://www.oracle.com/technetwork/articles/systems-hardware-architecture/ovm-blade-ref-conf-192667.pdf

Sun Blade 6000 I/O and Management Architecture:

• http://www.oracle.com/technetwork/articles/systems-hardware-architecture/sb6000-io-mgmt-400403.pdf

Sun virtual NEM Block Diagram: http://blogs.oracle.com/marchamilton/resource/hydra.jpg

Remora designing a complete red stack solution

Remora has been contracted to to provide architecture advice to develop a complete optimised solution for Oracle's eBusiness Suite, Oracle Fusion Middleware and Oracle Database Enterprise Edition.

Remora's team of engineers sized and delivered a complete solution using three SPARC Enterprise M4000 Servers for database workload. Oracle T-servers, including a T5440, two T5140's and one T5120 application servers were deployed for Oracle eBusiness and Oracle Fusion Middleware worloads.

Combining Oracle's server infrastructure and Oracle SPARC virtualisation technologies, a complete solution was provided for production, user acceptance testing and disaster recovery using Solaris Containers and Logical Domains.

It doesn't have to be hard

A common problem we've encountered is the need to replicate storage between sites for availability, performance or disaster recovery.

In the absence of any open standard for storage replication, this can be difficult to do across storage vendors and even across products from the same vendor.

It can be easily handled in sites which run Oracle Solaris, however, thanks to Storagetek Availability Suite. This particular solution can provide point in time snapshots and/or remote mirroring with _any_ storage connected to a solaris server. A link to information about this most interesting--and unique--product is here:

• http://www.oracle.com/us/products/servers-storage/storage/storage-software/034791.pdf

Other replication solutions available from Oracle work with storage devices from the same family--the Unified Storage or 7000 series, or the Pillar Axiom range of storage systems. Sun Cluster also supports vendor specific replication from Hitachi and EMC, as well as Oracle Data Guard and MySQL replication.

There are thus multiple ways to tackle this particular issue.

Remora rolling out an OVM cluster across 4 continents

Remora has been contracted to provide one of the largest Oracle Virtual Machine 3 deployments in the Asia-Pacfic region. OVM 3 is the latest virtualisation technology from Oracle and provides some major advancements from it's predecessor OVM 2.

Oracle Virtual Machine utilises the Xen hypervisor and is rapidly becoming the preferred option for VM deployments due to it's robustness, competitive licensing and support costs. When purchased with Oracle hardware it has no licensing costs and comes with 1 year support.

The deployment being rolled out for the large international client spans over 4 continents. Once completed, the client will be able to manage their entire infrastructure from a central location. The result will be a more robust global network with greater flexibility, high availability, greater redundancy and lower maintenance costs.

Simple, Reliable and Affordable

Oracle recently announced the Oracle Database Appliance, which offers small and mid-sized organisations a rapidly deployable, highly available platform at an affordable cost.

With this new offering, Oracle’s Engineered Systems family becomes more complete, with Oracle Exadata Database Machine supporting the needs of larger enterprises and Oracle Database Appliance supporting SMBs and departmental applications.

Showing the scalability of SPARC

Remora provided Oracle Enterprise M5000 Servers and Oracle Database Enterprise Edition for the provision of a human capital management system for over 15,000 employees.

Constraining Oracle Database licensing was key to the final solution, achieved with the use of Logical Domains to minimise licensing costs.

Installation and operational mentoring

Remora provided an installation service with a difference for Ford Motor Company of Australia, for one of their Oracle Sun M5000 servers.

In addition to physically installing the system into the rack, commissioning it and checking correct operation, we had the opportunity to provide Ford staff with on the spot training and mentoring in the various methods of administering Dynamic System Domains (DSD) on the server.

DSD allows authorised staff to re-allocate system resources--while they are live--to different workloads, called domains, inside the server. It is one of several virtualisation technologies provided in Oracle products.

Each Domain may be regarded as an independent box-within-a-box from both a hardware and software standpoint. DSD can be administered either by a secure web interface, or by the command line from a console session.

We even uncovered an undocumented feature of the interface on the day, and were able to provide Ford with additional information about it.

SPARC Enterprise M5000 Server

Oracle’s SPARC Enterprise M5000 midrange server delivers high scalability, mission-critical RAS, and advanced virtualization in a compact 10RU space. The ability to mix generations of processors provides excellent investment protection.

Make deployment of VMs easier

There are several reasons the use of templates in a virtual environment makes sense, they are quick to set up, taking mere minutes rather than an hour or so to bring up a fresh VM, and they are best used as 'Gold images', which are templates that are set up as close as possible to the best case for specific work loads.

When a template is used in a VM environment, they speed up the process of bringing up a VM, and help maintain consistency between several similar VMs.

By utilizing templates as 'Gold images', one can be assured that the VMs are capable of dealing with what they were designed to do, such as a LAMP (Linux, Apache, MySQL & PHP) template, which would make bringing up a webserver quick and easy.

Another benefit of templates, is they can be easily transferred between VM servers, and they are assured to be set up the same. Where that level of certainty cannot be given to freshly built VMs

Browse through the prebuilt templates available here:

• http://www.oracle.com/technetwork/server-storage/vm/templates-101937.html

JeOS creation is interesting

The issue with JeOS creation is, if you want to do it manually, it's difficult to actually find anything to help you.

Use this: http://oss.oracle.com/ol5/docs/modifyjeos.html
Or this: http://www.oracle.com/us/technologies/027000.pdf

To do it automatically download OVM template builder and basically just follow the prompts.

Highest Scalability, Performance, and Availability

Oracle's StorageTek SL8500 modular library system, the world’s first exabyte storage system, delivers unprecedented value through heterogeneous data consolidation and multi-generation media support in an ultra-dense footprint. With up to 100,000 slots and 1,000 PB (utilizing 2:1 compression). And it ensures that your data is safe and secure with 24x7 availability.

The World's Fastest Flash Database Accelerator

Oracle’s Sun Storage F5100 Flash array is the world's fastest and most power- efficient Flash array for accelerating databases and I/O intensive applications. It redefines database performance, cutting transaction times in half and doubling application throughput. The Sun Storage F5100 Flash array lets you scale your performance and capacity needs efficiently without impacting data availability.

Oracle’s Pillar Axiom 600 storage systems

Oracle’s Pillar Axiom 600 storage systems deliver industry-leading storage consolidation that combines high, predictable performance with extreme efficiency.