Oracle Engineered Systems for Dummies: for Decision Making [PART 4 & Final]

Oracle Engineered Systems

Click the following links to read previous parts of this series

Part 1

Part 2

Part 3

Oracle Engineered Systems as Simplified Cloud

The term “Simplified Cloud” is coined by Oracle earlier has been transformed recently to “on premise, cloud at customer and public cloud”. Essentially Oracle say that, its systems are extremely flexible in all these scenarios whether it is purchase and use on-premise , delivered to customer data center and managed by Oracle or use it from Oracle’s public cloud. Let us look on each of these options at a high level.

Engineered Systems : On – Premise

This case has been discussed in previous two parts of this series. This is about purchasing the OES  with a significant initial investment and using it at your data center and maintained by yourself or giving the maintenance to Oracle ACS. You should be aware of the business case of purchasing such a huge system and the roadmap of the business volume you are going to handle with this system. If you are small now and roadmap to huge growth is also on a slow progression, you should be extremely careful. On the contrary, if you are big fish with huge business volume and expect to grow exponential over coming two to five years down the line, this is a classic case to go with OES. Pls read through Part 1 and Part 2 of this series.

OES Cloud at Customer Model

This is a case where, the OES is delivered to Customer Data Center and/or wherever customer want that to be delivered (may be in a third party data center where customer pays the real estate fee) and completely maintained by Oracle. There are three typical business cases (where an assumption that customer doesn’t want to manage the system themselves)

  • The customer application can not afford to have application latency from the cloud and hence they want the database to be somewhere near the application proximity.
  • The data security and isolation is of utmost importance to customer and they don’t trust to host the same in cloud.
  • The demand is highly dynamic and fluctuating and hence it looks like a pay as you go model while the system sits very near to your application servers

In all these cases it is worthwhile to take care of the following.

  1. There is a minimum capacity available for OES. For example for Exadata, they deliver only Quarter rack. 1/8th rack is not available.
  2. Assess the workload for every month for at least 2-3 years and estimate how much compute is required for the same (Eg : in OCPUs; which is equal to one CPU in a Quarter Rack of Exadata. A Quarter rack contains 48 OCPUS). If this is really small , something like 6 OCPUs a month, then 2 OCPUs another month, then this model may be a more profitable one than purchasing a machine and managing it self.
  3. If the workload is consistently big which requires huge compute across the months / years to come, then , this model may not be a profitable proposition. See the following illustration.
    Suppose you take a Quarter rack of Exadata in this model which contains 48 OCPUs.

The list price of Exadata Quarter Rack     = $363K
Annual Premier Support fee                       =$73K
Oracle Exadata Software(for 36 disks)      =$360K
Annual Support for Software                      =$79K

If we assume 20% discount on list prices

The list price of Exadata Quarter Rack     = $254K
Annual Premier Support fee                       =$51K
Oracle Exadata Software(for 36 disks)      =$252K
Annual Support for Software                      =$55K

Assuming a 5 year amortization of capex, the annual cost will be
Annual Cost for Infra      = ($254K/5)+$51K+($252K/5)+$55K = $207K
Monthly cost will be $173K/12 = $17K

Oracle Exadata Quarter rack at Customer premise price is $13.5K Per month and $0.3226 Per OCPU. i.e for a Quarter Rack Exadata

Annual cost = =13500*12+ (72*0.3226*24*30*12) = $363K
Monthly cost will be $363K/12 = $30K

The annual cost difference is 43% higher in “Cloud at Customer” model.

This is how costs came out when you really look into the deep.  Pls note that, this is only Hardware piece (BYOL). If you purchase software clubbed compute the difference can come higher.

So, you should look into all permutation combinations before opting this approach. Volume is the key and if you don’t approach this strategically, you will be at loss.

OES in Public Cloud

  • This is the most marketed model by Oracle now where they offer full stack of compute , embedded license+ Compute and other services in Oracle public cloud. As mentioned in previous parts of this series and sections, if the business demands are extremely dynamic with lot of fluctuations up and down, Public Cloud will be a good approach. But again, do your homework well before opting OES in public cloud. Let us look at a rough illustration below.

The monthly flex cost of an Exadata Quarter Rack with 30 % discount applied since it is taken for an year) on BYOL,  is $475K.

i.e yearly cost of Exadata Quarter rack purchased by customer in their data center $207K Vs above $475K.

That’s a 56% difference!

Yes . it is!

That’s why I suggest, you study the workload conditions and roadmap extremely well and deep down. Oracle’s cloud needs to be touched with extreme caution. Following are the key points to note

  • If you go for Pay As You Go model (on demand) with huge compute and you remain there for some reason, the billing will be 40% more (on top of 56% mentioned above)!!. If you want to run it on PAYG model, make sure you state the business reason for it and ensure to update the senior management well upfront
  • Going on monthly flex with 4 years can bring down the above 56% difference to 44%. The difference will still be there. However, the entire stack is being the responsibility of Oracle including monitoring, patching, management etc whereas you need to manage everything yourself PLUS chase Oracle Support for maintaining it.
  • The message is clear. When you purchase OES, install it and configure it in your data center, there are some overheads on it including resources, Oracle Support overhead etc.
  • When in Oracle Cloud, the management is done by Oracle. The cost is obviously high. Its customer’s responsibility to assess the target workload and decide where it should go.  Needn’t say, how difficult is to move a workload from one location once it is LIVE. So, do the pre-work all by yourself and asses it correctly before taking this step.
  • For those customers who purchases Oracle Database PaaS in Cloud (which is Infrastructure + Licenses), the cost will be more higher. On the other hand, customers with some enterprise license agreement like ULA can benefit at this front.

Conclusions

  1. Oracle Engineered Systems will be profitable for an Organization with he target workload and an exponentially growing one at least 2-5 years down the line.  If the target workload is small or insignificant and/or with a weak roadmap of growth, think twice before opting Oracle Engineered Systems on-Premise.
  2. Provision costs for maintenance when you estimate overall management costs for OES. This will be significant. Before signing the contract for procuring OES, ensure that you engage senior people from Oracle Support and get their commitment to provide the support. This is a critical part missed by many organizations and will be a hurdle later.
  3. Five Nines of Availability (99.999%) can’t come organically with single box of an OES. Oracle’s Maximum Availability Architecture (MAA) is a costly affair. Opt it when the business demand it and create the right price for the end service you offer to customer to make up your costs.
  4. OES in Oracle Cloud may look attractive from a management perspective. But, it need to be adopted with utmost care , considering the nature of workload, significant fluctuations in usage, roadmap of business load, licenses in hand etc.
  5. Customers with Unlimited License Agreement (with an exclusive right use those licenses in Oracle Cloud) can benefit from Oracle BYOL model in Oracle Cloud. This is an important point to be considered during assessment.

Previous :

Part 1 – Oracle Engineered Systems for Dummies : for Decision Making [PART 1]
Part 2 – Oracle Engineered Systems for Dummies : for Decision Making [PART 2]
Part 3 – Oracle Engineered Systems for Dummies : for Decision Making [PART 3]

Written by
Sethunath U N

Note :
This is Part 4 and Final of 4 Parts in this series which discusses about Oracle Engineered Systems. If you think of considering Oracle Engineered Systems adoption or want to evaluate your decision, do contact us at mavisdx@mavisdx.com for a comprehensive analysis to help you take the right decision.

Leave a Reply

Your email address will not be published. Required fields are marked *

Scroll to top