Great product that could use some simpler VS setups
Updated April 20, 2020
Great product that could use some simpler VS setups
Score 6 out of 10
Vetted Review
Verified User
Overall Satisfaction with IBM Cloud Virtual Servers
It is mainly being used in data science and actuarial departments for heavyweight modelling where either high memory or large CPU requirements (or both) are required. Also as we explore further the use of machine learning and AI, the possibility to access very large GPU machines is appealing as this dramatically increases efficiency in a cost effective way.
- A broad range of setup options, so we can choose the right GPU/CPU/RAM for the run at hand, which makes things more cost-effective.
- Data security is important to us, being able to choose a server location helps with giving assurance that the data will be looked after appropriately.
- The cloud site is extremely well set out, it is very easy to quickly spec out and cost a VM.
- When launching a VSI it's not always obvious how far progressed or how long we will have to wait.
- It's also not 100% clear when billing starts and stops.
- It would be great to have a few "works out of the box" configurations, with sensible defaults so that inexperienced users can quickly launch a windows instance to run a heavy model without having to worry about getting config exactly right. "too many options".
- With access to large machines, we've been able to increase our modelling ambitions to include more complex assessment of the risk in the business.
- The ability to access large GPUs has singificantly cut the time it has taken us to train very complex AI systems
We don't use it, as our workloads tend to be time-critical and so we require on-demand dedicated servers.
During the last 6 weeks we have been required by our regulator to run our very large solvency models on a much more regular basis than normal. Being able to access reserved capacity has given us the flexibility to achieve that and impress our leadership team in the process! Which is a win for everyone
We have a very active and engaged IT security team and regularly need to process sensitive data. Being able to carefully control this aspect of hosting has made it much easier to get ITSec comfortable with using IBM Cloud VSIs
IBM, AWS and Google Cloud are all market leaders for a reason and in many ways there isn't much to choose - aside from personal preference of purchasing managers. In our case it was familiarity and confidence from having used IBM CVS on previous projects, the security/data considerations and wide ranges of configurations that swung the decision.
IBM Cloud Virtual Servers for VPC Feature Ratings
IBM Cloud Support and Implementation
For our more experience engineers the process was very smooth. There was an element of "lost in the vastness of choice" for our less experienced engineers which led to us setting up some standard configs to limit their choices. In general however the implementation is pretty good and easy to set up and monitor.
Premium - The processes we use CVS for are somewhat mission critical - as such spending that bit more on support was justified
In our case the best example of excellent support is in the excellent, detailled documentation that exists for everyone! Don't underestimate the time and effort it takes to design and put that all together!
- Central Documentation and Learning Hub
- IBM Cloud Docs
- Ability to check uptime status for IBM Cloud products
- Access to IBM and Developer communities through Stack Overflow or GitHub
Yes, in general by removing the guess work from set up, implementation and monitoring.
IBM Cloud Virtual Servers Pricing, Time savings and Efficiency
Whilst the 14% decrease was welcome, the main benefit was compared to hosting the hardware ourselves - the cost of which was very high given that we didn't fully utilise the processing power. Also as machine learning/AI has become more and more important to us - being able to access cutting edge GPUs cost-effectively (in an ever changing world!) has been very valuable too!
5-8 hours time saved per person per week
A good example of the flexibility/efficiency of hourly billing comes from urgent requests to run our very very large regulatory solvency model over the last few weeks (COVID19) to provide data to our regulator to assure them that we remain solvent. Whilst running the resources required are significant, but once finished the fact that we can "turn off" the resource means we really are only using it whilst needed.
Using IBM Cloud Virtual Servers
Actuarial modelling and calibration
Data science
Investment/asset management
Data science
Investment/asset management
3 - The internal support are basically passionate users, with massive hearts who willingly add to their busy workload by supporting others!
- Rapid scale up of processing resources to meet regulatory reporting requirements
- Very large RAM machines for large in-memory calculations (for example nested stochastic simulations)
- Large GPU machines for rapid training of complex ML/AI models
- A convincing Chatbot!!