Oracle Private Cloud Appliance Reviews

• PCA is flexible and supports different workloads, such as database, middleware, apps, etc.
• Supports many operating systems, such as most of Linux distributions, Solaris, and Microsoft Windows.  
• No software or firmware licences are required. All are included SW; free. 

• It is a converged system, which contains integrated HW components (compute servers, networks switches, and storage) and management software plus Oracle VM. 
• It provides more flexibility to choose the required OS. 
• It is scalable as the box can be increased by one compute node, if required. 
• It has helped us to reduce Oracle licensing costs for database and applications.         

• Scalability is a valuable feature, as compute nodes can be added as required. There is no force to increase form 1/8 Rack to 1/4 Rack as in Exalogic, for example. 
• All shipped SW is free, such as OVM and SDN. 
• Supports mixed compute node version, such as X4-2, X5-2 X6-2, etc., and capacity On Demand feature for both HW and Oracle licenses. 
• Fabric Interconnect Xsigo switches, which are used to connect PCA to the customer datacenter core switch or to the SAN switch, are very flexible and support all type of connectivity, such as QSFP, SAN port 16Gb, InfiniBand connectivity, etc.
• Supports connection to most of network topology, such as VLAN and network segmentation. 
• Oracle provides Linux and Solaris OS templates ready to create VM servers for PCA in OVA format.       

• OVM Manager Interface is using HTML5. It should be reconverted to be a Java interface. 
• A VM server console is needed as an improvement, especially when GUI mode is active.
• Oracle should provide the OVM client console per user based on his/her privileges to manage his dedicated resources and VM servers.
• Patching processing takes a long time. Only one compute node is patched at a time.
• Creating networks for PCA is done via CLI. It should moved to the OVM GUI with a clear diagram or graphics.
• The PCA dashboard needs more enhancement to contain more info about the components and showing the external connective network links.
• Internal ZFS storage at PCA is designed to be used for the OVM database, configurations data, and contain VM servers OS only. The customer has to attach an external storage from his own data (it is allowed to use non-Oracle storage), so this design is adding more complexity. I would prefer if Oracle provided another design or option with an internal scalable storage that can be used for both configurations data and customer data without the need to get external storage.           

Until now, it has been fine.

I have not scaled to test. 

I cannot give the support more than a six out of 10.

We switched to reduce Oracle licenses and support costs.

Initial setup was straightforward:

• Three management IPs form two OVM management nodes clustered in active/passive.
• Provision the compute nodes to OVM cluster.
• Connect the external storage. 

For normal setup, it will take around one day. 

We used a vendor team for the implementation. I would rate their expertise as a seven out of 10.

• No licenses for the shipped software or firmware.
• For pricing, it is more expensive than the normal solution or the hyper-converged system.  

VMware and Nutanix.

If you have many Oracle products, such as databases, middleware, and apps, it will help you reduce Oracle app licences and support costs. 

• Flexibility to run mixed workloads for Oracle and non-Oracle environments.
• One stop to all your applications; for example, consolidation of various platforms, Linux, RHEL, WM Windows and Solaris into one single system.
• Provisioning for rapid application deployments (in just a few hours).
• Also, cloud-ready solutions.

Any new deployment typically takes weeks or months to get things ready. With PCA, the deployment provisioning is made easy; the new environment can be ready in just few hours. Thus, this will hep organizations to avoid delays in launching new deployments.

• Support and more awareness of the system functionality and internal mechanism.

I have used it for less than six months.

Initial system planning is most important; how to distribute the resources, configuring the virtual network, and the best usage of OEM is important for the new deployments. Besides, teams with VM concepts and background would be handy to execute the project. When you have such teams available, you're unlikely to encounter any serious issues during the deployment.

Planning of resource distribution and management is a vital step in configuration; also, network setup. When you put best practices in place, you shouldn't be facing any stability issues.

We have not experienced any scalability issues yet.

Since the PCA is relative new, finding experienced resources is bit hard; need a lot of improvements in terms of documentation and technical details.

Technical support is satisfactory.

Engineered System setup is never going to be an easy or straightforward installation, anytime; requires solid planning and skilled resources.

A combination of Oracle Customer Support (ACS) and an Oracle global partner implemented it.

Rapid provisioning of new environments make launch of production on-time.

No additional software licenses are required for Oracle Private Cloud Appliance.
The Oracle Private Cloud Appliance system price includes all the required software. It's the most cost-effective solution for your mixed Oracle and non-Oracle workloads. Achieve efficiencies by only paying for the Oracle Database software licensing that you actually use.

We compared various cloud options that supports the full range of Oracle software, including AWS, MS Azure, etc.

Oracle PCA is definitely worth investing in.

It's considered a converged system, so one of the real interesting things about it, you can have mixed workloads. You can install web server, app server, databases, rack, all those things, so it's pretty much an open-ended environment where you can install pretty much anything that you want to install on it as it's x86.

One of the benefits, like I said, it's a converged system, so that I can install a web server and my entire EBS stock. I can have my app server, web server, my Oracle Database all on that one compute node, all on that compute stack, that is.

I think a tighter integration with external storage, the simplicity of having, let's say for example, EMC storage or Hitachi storage connected to PCA would have been a nice little added.

While it scales, the base configuration is two nodes and it can go up to 25, I believe, so it can scale as you need to. As your business grows, as your compute requirements grow, you just add nodes as needed.

The technical support's pretty good. Again, it's a fairly new product, but the OVM support team, the Linux team, as well as the traditional database support teams are all pretty good. We haven't had to tap into those guys too much but we know the times we did there were pretty good.

This is really a good consolidation play. So typically customers would have 10, 15 servers out there, probably older and what PCA does is give them a platform for consolidation. So it really gives them an area to consolidate not only the web servers, app servers, but database servers too.

Since it is a new product, there's a lot of new things we had to kind of learn, connectivity-wise, although it's based on Oracle Virtual Machine. So we knew that skill set before it came on so we had that a little bit going on. But some of the nuances of the new technology stack, the converged system, was something we had to adapt to and learn.

The key, first and foremost, is again, you have to have CAPEX or OPEX or both. So if that's your main driver is cost reduction and CAPEX and OPEX is one of those, then this is a good avenue to go to too solve those problems.

Hyper-converged and converged systems are the new thing that's happening in the industry and PCA just falls right into that category. There's a lot of options for customers to look at besides PCA.

Rating: since this is an early product, I think the good thing about it is it's moving up in the scale as opposed to moving flat. I'd probably give it a seven and a half and upwardly moving as it adopts to the industry standards.

My suggestion would be have a good organizational boundaries on how you're going to support this because it does tether around the borders of network, storage and database, so you have to have a good organizational definition of how you're going to support a configuration such as this.