Market Data as a Service
Sounds kind of cool doesn’t it? We’ve got everything else available as a service, anything from compute, storage, email, databases, software, platforms.. You name it, it’s available as a service, all except.. Market Data. But then, what do we mean by Market Data, certainly it’s not the generic reference to static market data, which is available across the board from various vendors, including aggregators like Bloomberg, Snowflake, SIX, ICE, LSEG, etc but let’s focus on the Real-Time (“I need it now”) access to financial content. Yes, typically this is a reference to streaming, ticking data, but it doesn’t always need to be, but the pattern helps understand the need.
Gaining access to streaming data typically involves building a MarketData Distribution System, an MDS. The responsibilities of the MDS are to ingest/acquire streaming data, and distribute it to applications (either user/display based or non-display). This all has to be done under the auspices of control. Building and owning the MDS means that you’ll need infrastructure, likely a suite of servers, and these servers will all need feeding and watering (power and cooling), so you’ll be renting part of a datacentre. In reality, to support any form of business continuity, you’ll need two datacentres and two sets of infrastructure/servers. That infrastructure is likely to be bare metal (physical machines) or it could be virtual machines, nonetheless, it needs to run an operating system, probably Linux. Then comes the software stack that runs on top of that infrastructure that delivers the MDS capability. Now, we need consider people that will ensure that infrastructure remains upto date, that security patches are applied, that has been configured, that runs smoothly, that provides business continuity, serves a growing business, et al.
Running an MDS, OnPremise (cloud or datacentre) is a factor of infrastructure, software and people. Together they form the total cost of doing business, in fact the Total Cost of Ownership. That TCO is only likely to move in one direction, upwards.
Infrastructure is highly sensitive to power consumption and the ensuing energy bills, is susceptible to increased compute capacity as time marches on and needs increase. At some point, technology refresh become an issue where the capital purchase needs to take place again, to replace elements of the infrastructure. One client we worked with had in excess of a 100 servers for their MDS, and the technology refresh was circa $1m, even though this event had been moved out from 5 years to 7 years.
The software stack is complex challenge. It needs constant updates, not only to avail of new features and fixes, but also to keep on top of vulnerabilities. This applies to the MDS software, the operating system, etc. For each software component to need to ensure you have full cover in terms of support, maintain your relationship with each of the vendors. Even with all of that, is there a risk of breach, have you been independently assessed? If your MDS is in a datacentre, the risk of breach is probably far lower, but with cloud one needs to consider very strict (least privilege) processes and configuration. One thing not mentioned yet, which was alluded to at the beginning, is that to operate the MDS, you’ll need to do with governance and control and by that we mean Entitlement and Reporting. This key critical component is the reference point for supporting billing, auditing and business enablement.
People, whilst immensely valuable with their collective expertise and skills, become a risk, the key person risk. Technologists attribute value to their skills and expertise, and whilst looking to broaden their CVs will always consider improved offers from other organisations. Thereby lies the challenge of retaining key persons, to ensure they have the job satisfaction and personal rewards, along with the financial compensation. This applies across the board to those with specific MDS skills and experience, to those with specific skills around cloud technologies. One trend suggests that the pool of MDS skills and experience is diminishing.
The thought for your organisation is, are your strengths and value best served by running and operating the technology, or is it, ensuring the business has the data, in a timely fashion, to be efficient and competitive. let’s examine the financial argument where it might be time to consider an alternative.
This chart is looking to highlight that; an OnPremise/Cloud deployed MarketData Distribution System (MDS) is going to attract the costs that were discussed earlier. The horizontal blue line represents this TCO. In reality the TCO is probably going to take step jumps as you incur technology refresh, or, need to scale. There’s not many circumstances where this line is going fall. To make a decision around which option to consider, look at the point of inflection. If you’re to the right of it then the choice of a deployed MarketData Distribution System is probably the right decision. However, if you’re to the left of the point of inflection, this would certainly be a reason to consider MarketData as a Service as an alternative.
Previously, we spoke about Bloomberg’s Enterprise offering of B-PIPE. This offering offers up one or more Delivery Points as one or more IP addresses. Everything behind these IP addresses is owned and operated by Bloomberg. An organisation’s responsibilities stop at this Delivery Point boundary. There are nominal costs for the Delivery Points with availability for; OnPremise (in a datacentre), via the Internet, and, full support in Azure, AWS and GCP. Additionally, the billing/costs are based on usage, so very cloud-like elastic billing. You branch into new products, new clients, new markets, your usage and costs increase. Likewise, if you pull out of a market, your usage decreases, then so do your costs. This is largely measured on a monthly basis.
However, whilst a MarketData Distribution System largely acquires real-time ticking streaming data and distributes it (fan out) onwards to applications, Bloomberg’s B-PIPE provides a full ecosystem of micro-services that goes way beyond the streaming data. Examples include, real-time access to static/reference data, access to level 2/3 full-depth order books, ability to submit/route order into EMSX, submission of pricing into Bloomberg’s OTC venue (ALLQ/TOMS), access to historical data, machine readable news, access to MARS capability, and much more. The biggest benefit comes when you have a population of Bloomberg Terminals as these users will attract significant cost savings, further leveraging your Bloomberg investment.
If you’re interested in examining whether MarketData as a Service is a good fit, looking for assistance in building a business case to make an informed decision, or, just want to understand the implications of migrating from an MDS to MDaaS, please don’t hesitate in getting in touch with our experts at Sherpa.
