Oracle Exalogic [EOL] Reviews

We currently use Exalogic for SAP solutions. We also use their Exadata solution for the data warehouse for the Oracle database.

The performance is the solution's most valuable aspect.

The solution needs to add a network virtualization feature similar to that of VxRail. If they could implement that, it would be great for the product.

I've been dealing with the solution since 2014.

The solution is stable, but it typically takes time to build up to stability. Sometimes there are some bugs. Overall, we find it to be quite reliable.

In the context of ACI, it is scalable. You can add anything from a quarter rack or half rack to a full rack, etc. It is scalable in terms of computing and in terms of storage as well. However, if you think about ACI, Hyper-converged network and some of the solutions like VxFlex and software-defined storage and VMware, it's not that flexible. The level of scalability depends on the conversion with other products.

Typically, if a client needs support, they contact us and we are the ones to offer it to them.

However, we have contacted Oracle technical support and we have a back-to-back contract with them. They support our HR on-site team. We also get support from Oracle functional services and backend support. They are usually quite good, but at times it depends on which engineer is working. When we get support from India, there's sometimes a delay. If it's from Europe or the US, often we get faster results. Time zones also affect technical support response times and will dictate who responds to our requests.

We're using a combination of products, including SPOC systems, Exodus servers, etc.

The initial setup isn't typically that complex.

On average, installation takes about seven days. This includes setting everything correctly.

Once it has been implemented, a company only really needs two people to maintain it. Usually, this is done by a database team and/or platform team.

We get the assistance of the vendor for the implementation. Oracle ACS helps us with the setup.

I understand that the licensing of the solution is pretty expensive, although I don't have exact numbers. The pricing depends on how much hardware, and if we're using professional services, etc. It's not just the license; there are additional costs.

We're an Oracle partner.

Typically we implement this solution on a large scale to enterprise-level clients. Typical customers might be big banks in Bangladesh.

I'd recommend the solution, but there are other options that give a client more flexibility. 

I'd rate the solution eight out of ten.

The database is the solution's most valuable feature. Running the application on one machine makes it easier to position it to the client. I don't need a separate machine to run my databases. I can virtually run the databases with the application. The Linux component of it is great as well. It makes it easier to administrate. There are a lot more Linux administrators as opposed to the SuperCluster which uses Solaris and there are not as many people that use Solaris.

The automation and provisioning of the environment, as well as the centralized monitoring, have all improved over the years.

The solution is prone to disk failures. This needs to improve. Over the past four years we've had to replace about six of them. They've failed almost every three to six months. 

The solution would benefit from the addition of real-time physical monitoring, like noting the temperature of the actual physical machine. The console should be able to tell you if it's overheating right away. Right now, it's not really in real-time. If I take my own temperature measurement versus what the console tells me, I find that my readings are more accurate than on the console itself.

The stability is very good. Once you configure it, it runs with no issues. Out of ten I'd give it an eight.

Scalability is quite easy. We are running two of them across two data centers, so it's quite easy to scale up and set it up. I'd rate scalability ten out of ten.

I'd give technical support a three out of five. We need to go through technical support when we experience a disk failure, which happens quite often. Once they provided the wrong disk size. We had to return them. We had some downtime because of this, whereas in the product brochure they said there's literally zero downtime. The delivery time for the disk took too long. We had to wait for more than two weeks for the disk to be delivered. Due to this hold-up, we have had more downtime than we had anticipated.

Previous to this solution, we used the Oracle SuperCluster. That solution was much more stable, but also much more expensive. The biggest issue was Solaris, the processor. There's no skillset for Solaris. Also, some of the libraries that are needed, you can only get them from Oracle. With Linux, it's generic enough that you can build solutions and you can get the libraries anyway because Linux is open source.

The initial setup was straightforward.

We had someone come and assist with the implementation.

I'm an Oracle partner. We use the on-premises deployment model.

I'd rate the solution eight out of ten.

This product offers a strong infrastructure with compute, storage, and network in a private cloud environment, enabling the organization to concentrate on business application development and improvement.

• High-speed network
• Strong storage features (ZFS)
• High Availability out of the box
• Integrated with a large number of products

With the shift to public cloud, and with Oracle releasing Oracle Public Cloud Machine and Oracle Private Cloud Appliance, the Exalogic X6-2 version was still released, but, clearer goals in the product roadmap are necessary to be able to affirm the improvements.

Initially, yes, there were stability issues. A number of redesigns were needed to stabilize the Hypervisor, control the stack layer, as well as the software defined network drivers (Mellanox bugs causing compute nodes to reboot). The latter was addressed by a heavy work initiative, but required a set of patching, impacting customers' system stability.

Just cost. To expand the hardware's capability requires a massive investment.

Depends on the level paid for. The Platinum Services offered by Oracle's Advanced Customer Services offers a multi-level mechanism of automatic service request creation and advanced monitoring. When all of its support features are integrated with customer's incident and change-management systems, it is a very effective model, with a proactive hardware replacement component. In cases of hardware fault, there is automatic shipment of parts without much effort to get it quickly fixed.

VMware. We switched because of the capabilities offered with Exalogic for integration with other engineered systems, which increases the throughput capability and enables processing of bigger and heavier workloads.

The initial setup is straightforward, but post-setup requires a very detailed and accurate design/architecture to avoid re-work in the storage and network layers.

Pricing and licensing are two very delicate aspects when talking about Oracle, especially engineered systems. The prices are quite high compared with other similar products, so the CapEx analysis and investment return have to be very well studied.

I rate this product eight out of 10. After years of stabilization (a lot of known bugs and issues with compute nodes rebooting due to InfiniBand HCAs and driver bugs, etc.) the product has achieved stability as a private cloud appliance.

Make sure you align the application and integration architecture with the sizing and the product's distribution. As its virtual datacenter is oriented to the network structure, once the VMs are created with an initial network configuration, it is not possible to change, despite destroying and recreating the VMs. This is acceptable in some cases, as long as the remaining automation process can reset the VMs.

decrease the licenses cost of oracle apps and FMW. AS Exalogic support oracle trusted partitioning license policy to license oracle apps.  

It has improved the way my organization functions by migrating apps to one consolidated platform that is dedicated to WebLogic and Oracle apps.

• Good performance : Exalogic is designed especially for Oracle middleware and applications hosted on WebLogic servers or any other application running on the top of Linux/Solaris OS, Exalogic HW is integrated with Elastic cloud software that performance optimization and good performance for Fusion Middleware modules. Also, Infiniband switches add high I/O throughput.
• Good platform for building easy clusters of vServers through ZFS storage: Exalogic has internal ZFS-ES storage which is supporting NFS sharing. Through NFS, WebLogic cluster (based on shared files system) can be implemented in an easy manner.

• EMOC needs more attention and simplification.
• The integration with EM cloud manager should be straightforward.
• The ZFS interface (BUI) should be simplified.
• Modifying the OS template should be more flexible.
• Backup and restore using the Exabar tool should be redeveloped.
• You can not restore a deleted VM directly; you have to create a VM and then restore to it.
• You have a limitation with creating attached volumes to the VM (just 4 volumes including the template volume).
• The included ZFS doesn't support replicating data inside the local disk (attached volume). You have to convert the local storage (attached volume) to shared storage NFS.
• It doesn't support OVA.
• You can not share the internal ZFS storage with any platform other than Exalogic, so it is a closed box.

We have encountered stability issues: a storage failure that led to Exalogic being fully down' sometimes the storage cluster OCFS2 used for the OVM repository had stopped working and could not be mounted; finally, after many months, Oracle support discovered problems with InfiniBand (IB) host channel adapters (HCAs) that connect ZFS storage to the hosts and many firmware upgrades.

Sometimes while restarting the VM servers, they get stuck or no NIC is attached to the VM.

We stopped the update phase for adding more resources based on a management decision.

Customer Service:

Customer service is 5/10.

Technical Support:

Technical support is 5/10.

We previously used a different solution; we switched to get one platform for WebLogic-based apps.

Initial setup is complex and you have to fill in a very complicated sheet.

Implementation was done by the vendor team. Their level of expertise was about 5/10.

For pricing and license, you have to pay for Exalogic HW plus elastic cloud software for all cores at all nodes, so it will be expensive, including when upgrading. For example, to upgrade from 1/8 rack to 1/4 rack, you have to pay for 144 CPUs (144 * .05 (core factor for Intel CPU) * 10000 (price list for elastic cloud software )) + 22% for support per year + 130000 (price list HW upgrade) + 15600 HW support per year + services.

But Exalogic will help save on MW licences, which are more expensive than HW, as described elsewhere.

Before choosing, we evaluated other options; normal HW.

The best thing from my point of view for Exalogic is that it enables you to control Oracle middleware licenses and apps, which are very expensive. It supports a trusted partition policy for licenses, so Exalogic is helping to reduce Oracle MW and apps licenses, as you are licensed only per the physical cores assigned to the VM server only and you can increase the licenses by increments of 2, 4, 6, etc. physical cores. You can not find this feature outside Oracle HW such as Exadata, Exalogic, PCA, etc. For normal physical HW, you have to pay for all cores in the physical sever, even though you might have a small load. For VMware, it will be a big problem and a licensing nightmare, as you have to pay for all of the VMware CPU cores in the cluster, even though you might only be using 4 cores!!!!! It is an unfair policy.

Exalogic is not easy for administration.

Internally, server consolidation. In a cloud architecture, moving the HA VM from one to other is faster. This is because of the InfiniBand network fabric between the compute nodes and storage nodes (ZFS Storage). It's easy to achieve the goal of always available functionality.

Externally, interoperability with data warehouse, BI or a big data environment.

Scalablility: When we need to upgrade from an eighth-rack to a full-rack, the real downtime is less then an hour.

Manageability: When we deploy the solution for a customer, we generally need to integrate with an existing network. With the Exalogic Configuration Utility we can provide either as an isolated or integrated network segment.

Enterprise high-speed interoperability: With IPoIB we can deliver to other environments, either IP-Ethernet-based or IP-InfiniBand-based protocol, both with minimum bandwidth capacity of 10Gbps, 20Gbps, or 40Gbps, redundantly.

• Highspeed transactions per second for business areas such as public-service, telco, banking, online store, payment gateway).
• Middleware application service, suchas statistics, data mining, could be better.

No stability issues.

No scalability issues.

Seven out of 10.

Yes. We switched because of

• highspeed InfiniBand fabric interconnect
• proper connectivity with Oracle Data Warehouse, OLTP, or OLAP.

Straightforward when we can isolate the network segment to be integrated. Possibly complex when we don't have any information about an existing network.

Get the price of the license you will need for the middleware that will be running on top of the Exalogic. Otherwise, you can use another middleware solution, as long as it can run on top of Oracle Linux.

Yes. VMware ESXi with commodity hardware, and Huawei FusionCube.

Before implementing make sure you have network planning confirmation and existing network services, such as DNS and NTP server.

The ability to have ZFS (shared file systems) with InfiniBand, and an InfiniBand connection to the Exadata database is very useful.

We used it create a new filesystem, and a stretched cluster with a few simple and quick steps.

Whenever there is a problem, opening a service request to get help from Oracle is an endless story.

For example, I had problems that I lived with for eight months. The answer that goes out of fashion is "the latest fw", but updating it also requires a downtime of 48 hours. It doesn’t mention when these interventions fail and must be rescheduled.

I have used Oracle Exalogic for two years.

To be honest, there was a stability issue. As I mentioned, we had to live with some bugs for months.

Scaling up the server required long machine stops of 24 to 48 hours. The interventions did not always end successfully. Sometimes they failed and needed to be rescheduled.

Technical support makes my life difficult.

The purchasing department chose this solution. Economically speaking, it is probably a cheaper option than others available, but for the technicians who work with this machine, it is fine as long as things are going OK; but every problem requires months to resolve.

The initial configuration is complicated. They deliver to you several sheets on which you need to declare, for example, a huge amount of VLAN and IP addresses. To make matters worse, the help they give you to understand exactly what you're setting is ridiculous.

We implemented it with the vendor, of course; but again, it would be nice to have more professional explanations.

My advice is to evaluate other solutions!!!

Storage. Strong in data integrity.

It doesn’t improve the way my organization functions. It has only one feature that improves app performance, but in a specific way. Everything else can be done with any other infrastructure hardware/software.

- Management console: Awful to use it, complicated, too many options spread around many menus that don’t make any sense.
- Networking: Everything is locked. It doesn’t allow you to change an IP address after creating a VM.
- VM template creation: Too complicated. Better install manually.

I’ve been using it for 8 months.

I encountered deployment failures for a simple VM (disk problems). Every deployment or configuration change in Exalogic is a surprise. It is not reliable.

However, once it’s running, it does not stop.

I’d say technical support is 9/10, but you need to keep up the pressure or else it will be “forgotten”.

I previously used a different solution.

I did not choose this product. I run it for a customer in his environment.

I wasn’t part of the initial setup. I can’t give my opinion.

Oracle implemented it.

I’m an outsourced provider for this client/product. I don’t have financial numbers regarding ROI.

I’d go for this product only if I have a really specific application to run, and it has proven better use with this product. Otherwise, I’d look for anything else.

The most valuable features are the Oracle Enterprises Manager Ops Center (EMOC) and the Data Centre Segmentation through Accounts.

This product has improved our organization by giving us better performance on WebLogic-based applications.

I would like to see more than four virtual disk capability in future versions of this product.

We have used this product for three years.

We received a high level of customer service.

We evaluated other solutions and chose this one because of the lower cost.

Initial setup was complex in the sense that it was a new deployment and it involved three Exalogic appliances.

We implemented this product through a vendor team.

Eliminate EMOC annual licensing. It should be free as an OVM.

It is good for Oracle in an Oracle environment.

The most valuable features are a fast and effective UI, and easy provisioning. Automated tasks can be instituted during service delivery. I use the available templates for PeopleSoft and ESB.

We use this platform internally for presales demonstrations, proof of concepts, training, and hands-on labs with our customers.

• Expand the templates and appliances like those that are available in OVM 3.4, such as the Oracle database firewall used in IPoIB private networks.
• Use of Virtual Network Services (VNS 1..02) for multi-tenancy (MT) use cases, and building secure enclaves.
• Continued evolution of the integration with EM 13c to extend MWaaS/PaaS and IaaS on Exalogic into the Oracle Cloud.
• Tell the great story of WebLogic Partitions for separation and promotion up/from the Oracle public cloud (using Exalogic as the on-premise platform of choice).
• Using OVAB or something like that to support Exalogic would be nice to have. The OVAB studio is not supported on Exalogic. That nurtured a service opportunity for application deployments using OVAB on Exalogic.
• Setting up different InfiniBand partitions in the EMOC would be a nice to have for additional MTs.

I have used it for three years.

The root filesystem on the physical computer nodes began to fill-up as images were added with each quarterly PSU upgrade. We might want to expand that filesystem to account for that. Also, provisioning more than a dozen VMs can be problematic for the proxy servers to handle, and some of the jobs fail. There seems to be no issue with less than 10 VMs.

There were no scalability issues to date.

We received very good technical support.

We had previous experience with HP’s CloudSystem Matrix and CloudSystem Enterprise. The Oracle platform has a much better and tighter integration between the network, storage and software.

The setup was straightforward using the ECU.

We did not evaluate any other options before choosing Exalogic.

If you are using only LINUX/Solaris and Oracle Fusion MW products, then this is a great platform.

The most valuable feature I've seen in the Exalogic platform is the integration it has with the Exadata. When I'm running WebLogic based wordloads, that really need low latency to an Exadata for improved input, improved latency, that's the big value of the Exalogic.

The benefits is it reduces the latency, reduces the amount of time for the Exalogic to get data from the database on the Exadata. When you're running OLTP applications that are very time sensitive, that's where it really comes in to play because you see the performance noticeably improve compared to other solutions.

There's a lot of additional features I would like to see in the Exalogic. The user interface would be nice to better integrate into Enterprise Manager 13 and eliminate the whole EMOC layer to the Exalogic. That causes a lot of confusion both internally as well as clients that use the platform.

I have been working with Exalogic systems for the last three years

A few, mainly due to how the OS deals with infinniband. Once you learn how that works your fine.

Stability of Exalogic's been great. In a virtualized environment, runs OVM, doing a lot of work with it. In fact, we've been using some of the Exalogic here at this event to run some of our hands-on labs that I've been doing as community efforts and haven't had any issues with stability on it, no big panics. With any system, you do have to patch it. If you don't patch it for a year or two, you might start getting some bugs.

Scalability, you can take it up to a full cabinet and that's not been an issue at all. It's a lot of horsepower in a single cabinet.

Technical support for Exalogic is okay. It's acceptable. Again, it's a complicated platform and it has some limitations that tend to have to call support a little more often.

What you really look at on an Exalogic is what's your application and would the benefit of the lower latency when you connect to the Exadata really improve your application to the point that that's the solution you want to go to.

The initial setup on the Exalogic when you're running it is pretty straight forward. It's an engineered system so most of it's up and running. You put it in the data center, do some configuration on it, and you're up and running in a day.

Rating: I'd give Exalogic probably a 7.5. The reason is, the issues with the Emoc system and that integration, that user interface is very different. Also, some of the limitations on the Exalogic. You have to use Oracle's OS load, you can't run Windows. There's some limitations to how it works that causes some confusion. That's why I'd rate it a little lower.

My recommendation for peers if you're looking for a need to reduce the latency for your workload, definitely take a look at it. It's definitely going to be the fastest thing out there to solve that issue for your connectivity to the Exadata when you're running a WebLogic based application. You need to take a look at it for that reason alone. Be prepared for some of the limitations and make sure your application works within those limitations.