Caffe Deep Learning Framework vs. Oracle Database

Caffe is only appropriate for some new beginners who don't want to write any lines of code, just want to use existing models for image recognition, or have some taste of the so-called Deep Learning.

Incentivized

We migrated from NoSQL to an Oracle database. One of the reasons was robust backup and recovery options available in the Oracle database, which provide zero data loss. A transactional database like Oracle is a better fit for our use case than NoSQL. On a large scale, deployment was evaluated as a cheaper option than the NoSQL engine. This conclusion came even after considering Oracle license is expensive.

Incentivized

• Caffe is good for traditional image-based CNN as this was its original purpose.

Incentivized

• Supports most of the Operating Systems like Unix, Linux and Windows Server.
• It works well in high load environment under intense parallel transactions setup.
• Highly reliable DBMS, especially RAC is very much reliable.
• Well managed and predictable release of security patches.
• We have highly scaled it from on-prem to a cloud cluster environment for our product.
• One of the best-performing DBMSs on Linux machines under test delivers high throughput (QPS).

Incentivized

• Caffe's model definition - static configuration files are really painful. Maintaining big configuration files with so many parameters and details of many layers can be a really challenging task.
• Besides imagine and vision (CNN), Caffe also gradually adds some other NN architecture support. It doesn't play well in a recurrent domain, so we have to say variety is a problem.
• Caffe's deployment for production is not easy. The community support and project development all mean it is almost fading out of the market.
• The learning curve is quite steep. Although TensorFlow's is not easy to master either, the reward for Caffe is much less than the TensorFlow can offer.

Incentivized

• The memory demand and management makes it impossible to run it in a container.
• It is hard to perform local unit testing with Oracle even using the personal edition (aggressive all the available memory grab for itself).
• Lack of built in database migrations (e.g. as Flyway).
• The need to install the Oracle client in addition to its drivers.
• The cost of running it, especially in the Cloud.
• Comes with very spartan community grade client/management tools whereas the commercial offerings tend to demand a premium price.

Incentivized

There is a lot of sunk cost in a product like Oracle 12c. It is doing a great job, it would not provide us much benefit to switch to another product even if it did the same thing due to the work involved in making such a switch. It would not be cost effective.

Incentivized

Many of the powerful options can be auto-configured but there are still many things to take into account at the moment of installing and configuring an Oracle Database, compared with SQL Server or other databases. At the same time, that extra complexity allows for detailed configuration and guarantees performance, scalability, availability and security.

Incentivized

1. I have very good experience with Oracle Database support team. Oracle support team has pool of talented Oracle Analyst resources in different regions. To name a few regions - EMEA, Asia, USA(EST, MST, PST), Australia. Their support staffs are very supportive, well trained, and customer focused. Whenever I open Oracle Sev1 SR(service request), I always get prompt update on my case timely. 2. Oracle has zoom call and chat session option linked to Oracle SR. Whenever you are in Oracle portal - you can chat with the Oracle Analyst who is working on your case. You can request for Oracle zoom call thru which you can share the your problem server screen in no time. This is very nice as it saves lot of time and energy in case you have to follow up with oracle support for your case. 3.Oracle has excellent knowledge base in which all the customer databases critical problems and their solutions are well documented. It is very easy to follow without consulting to support team at first.

Incentivized

Overall the implementation went very well and after that everything came out as expected - in terms of performance and scalability. People should always install and upgrade a stable version for production with the latest patch set updates, test properly as much as possible, and should have a backup plan if anything unexpected happens

TensorFlow is kind of low-level API most suited for those developers who like to control the details, while Keras provides some kind of high-level API for those users who want to boost their project or experiment by reusing most of the existing architecture or models and the accumulated best practice. However, Caffe isn't like either of them so the position for the user is kind of embarrassing.

Incentivized

Because of a rich user base and support for any critical issue, this is one of the best options to choose. In case the project has a TCO issue, it can compromise and choose Postgres as the best alternative. SQL server is also good and easy to code and maintain but performance is not as good as the Oracle

Incentivized

• Since we stopped using Caffe before it can reach the production phase, there is no clear ROI that can be defined.

Incentivized

• Multiple applications can use the same database and still get high performance
• Licensing cost is still a concern compared to the other options available in the market that are very very inexpensive
• Almost a maintenance free database
• Oracle Grid makes life easy in terms of monitoring and managing the databases

Incentivized