Trading Desk IT Setup in Dubai: Redundancy, Records and the Payment Fraud Problem
A trading desk fails differently from an office. What to build for redundancy, records and the payment fraud aimed at firms that move large sums.
A trading desk needs three things an ordinary office does not: connectivity that genuinely fails over rather than nominally, a decision about which communication channels are business records, and a documented procedure for verifying payment instructions. The third is the one that prevents the largest single loss most trading firms will ever face.
Key takeaways
- Two lines are only redundancy if they follow different physical routes.
- Cable every trading position; wireless belongs in meeting rooms.
- Decide which channels are business records before you need to produce one.
- Retention defaults are far shorter than the time a dispute takes to surface.
- Telephone verification of changed bank details stops the attack MFA cannot.
- Small headcount plus high value means key-person risk is the hidden exposure.
Redundancy that is real rather than nominal
Most offices that believe they have redundant internet do not. Two lines from two providers can still share the same physical route into the building, the same riser, and sometimes the same street-level duct. When a contractor puts a digger through that duct, both lines fail together and the firm discovers the arrangement it paid for was never redundancy at all.
Real redundancy requires asking the awkward question of each provider: by what physical path does this circuit reach the building, and does it share anything with the other one. In Dubai towers the answer is frequently that both enter through the same riser, at which point the honest options are to accept the residual risk or add a cellular link as a genuinely separate path.
The failover itself belongs on the firewall, configured and tested, not left as a manual process somebody performs under pressure. Test it by unplugging the primary during a quiet period and watching what actually happens to a live session, because a failover that takes four minutes and drops every connection is a different product from one that takes fifteen seconds.
Write down what the tested failover actually did, including how long it took and what broke. That record is what lets you decide whether the arrangement is adequate, and it is the document that settles the argument later when somebody asks whether the second line was ever proven to work.
- Ask each provider for the physical route, not just the service name.
- Treat a cellular link as the genuinely diverse path where risers are shared.
- Configure failover on the firewall; never rely on a manual switch.
- Test by unplugging the primary and observing a live session.
Why the desk should be cabled
Wireless has improved to the point where most offices run on it without complaint, and trading desks should still be cabled. The reason is not average speed but consistency. Wireless shares airtime with every other device in range, including the residential and neighbouring-office networks around a Dubai tower, and its performance varies minute to minute in ways nobody can control.
For a desk where a short interruption during a price movement has direct cost, that variability is the problem rather than the throughput. A cabled connection to each trading position gives predictable behaviour and removes an entire category of intermittent fault that is genuinely difficult to diagnose after the event.
Keep wireless for meeting rooms, visitors and staff moving around the floor. Our office network setup page covers the cabling and switching side, and why office Wi-Fi is slow covers what makes wireless unpredictable in towers.
Power, which is usually forgotten
A trading desk depends on more than the network. The comms cabinet, the switches, the firewall and the desk equipment itself all need power that survives the short interruptions buildings routinely experience. A brief dip that a laptop rides out on battery will still take down a desktop, a monitor array and a switch.
Uninterruptible power on the cabinet and on the desk positions turns a building power event from an outage into a flicker. This is inexpensive relative to what it protects and is frequently absent even in well-equipped offices, because power is assumed rather than designed.
Size it for a clean shutdown and a short ride-through rather than for hours of operation. The objective is to survive the interruption and the switchover, not to trade through a building-wide failure.
Check what is actually plugged into the protected supply after any desk move. Power protection is commonly installed correctly and then quietly undone months later, when somebody adds a monitor or a heater to the nearest available socket and finds it was the one feeding the cabinet.
Deciding what is a business record
Trading conversations spread across email, chat, voice and messaging apps, and the drift is towards whatever is fastest in the moment. That is efficient until a counterparty disputes what was agreed, at which point the conversation that settles it is on somebody's personal phone and is not retrievable.
The decision that has to be made, by someone senior rather than by IT, is which channels are for business. Once that is decided, those channels are captured and retained and the others are explicitly not for deals. A firm that has made this decision can produce a record; a firm that has not will discover the gap at the worst moment.
Retention then needs setting deliberately. Platform defaults are typically far shorter than the time a commercial dispute takes to surface, and a mailbox that quietly disposes of items after a year will not help in a disagreement about a cargo lifted eighteen months ago. Microsoft sets out the mechanics in its guidance on how retention works.
Tell the desk what was decided and why. A retention policy nobody has explained is experienced as arbitrary deletion, and traders who do not trust the official channel will quietly move the conversation somewhere they consider safer, which recreates precisely the problem the policy was meant to solve.
The attack aimed specifically at firms like yours
Business email compromise targets organisations that move large sums quickly against emailed instructions, which describes a commodity trading firm precisely. The attack is patient rather than technical. An attacker gains access to a mailbox, usually through a reused or phished password, and then reads quietly for a fortnight or more, learning counterparties, payment patterns, tone and timing.
When a genuine invoice is due, they alter the bank details, often intercepting and editing a real thread so the message arrives in a conversation the recipient already trusts. The request arrives at a plausible moment, from a plausible address, about a payment that was genuinely expected. Nothing looks wrong, because almost nothing is wrong except the account number.
Two controls defeat it, and they work at different points. Multi-factor authentication on every mailbox prevents the initial compromise, and the CIS Critical Security Controls treat it as baseline. Then a documented rule that any change to counterparty bank details is verified by telephone, to a number already held rather than one supplied in the email, stops the payment even if the mailbox was read. The second control costs nothing and is the one that survives a successful compromise.
Rehearse it once. Ask a member of the finance team what they would do if a counterparty emailed new bank details an hour before a payment deadline. If the answer involves replying to the email to check, the procedure exists on paper only, and that is the moment to fix it rather than after a transfer has left.
Key-person risk in a small firm
Trading firms concentrate value in few people, and IT usually concentrates the same way. One person holds the administrator passwords, understands how the systems fit together, and is the reason things work. That is efficient until they are unreachable, and the exposure is disproportionate because the business cannot pause while it works things out.
The mitigations are dull and cheap. Administrator access separated from daily accounts and held by more than one person. Credentials in a company password manager rather than a personal one. Documentation of the environment, so the knowledge is not resident in somebody's memory. And a named external provider who already knows the setup, so there is somebody to call who does not need a week to orient themselves.
For a firm where a single day of disruption can exceed a year of IT spending, this is the cheapest insurance on the list, and it is the item most often deferred.
Test the assumption rather than trusting it. Ask what would happen this week if that person were unreachable for three days: who has the firewall password, who can add a user, who can call the internet provider as an authorised contact. The answers are usually worse than expected, and the fix is an afternoon of documentation.
Supporting a desk that starts before the office does
Trading hours are not office hours. A desk covering European or Asian sessions starts early, and a problem at seven in the morning cannot wait for a nine o'clock helpdesk. Support arrangements written for an office week are a poor fit and tend to be discovered as such during the first early-morning incident.
What works is covered hours set around when the desk actually trades, with maintenance and updates scheduled firmly outside them. Patching a firewall during a session is the kind of avoidable self-inflicted outage that damages confidence in IT for years afterwards.
Monitoring matters more here than in most offices too, because the first sign of a degrading line should be an alert rather than a trader noticing that quotes are stale. Our IT support for energy and commodity trading firms page covers how we set the agreement around a desk, and managed IT support covers what the monitoring includes.
Trading desk IT setup: a sensible order to build in
If a desk is being set up or rebuilt, the sequence that avoids rework is: establish genuinely diverse connectivity first, because it constrains everything else and has the longest lead time. Then cable the positions and fit power protection, because both are disruptive to retrofit once people are trading.
Next, the security baseline, with multi-factor authentication enforced and administrator access separated, before the desk holds anything worth taking. Then the records decision and its retention configuration, which is a business conversation that often takes longer than the technical work. Finally backup and monitoring, with a restore actually tested rather than assumed.
Done in that order the desk is protected before it is valuable. Done in reverse, which is common, the firm spends its first busy quarter discovering which parts were skipped.
Frequently asked questions
Ask each provider for the physical route into the building rather than accepting the service description. In Dubai towers both circuits frequently share a riser or a street duct, which means one contractor's digger takes out both. Where the routes are shared, a cellular link is often the only genuinely separate path available.
It is good enough on average and unpredictable minute to minute, which is the wrong trade for a desk. Wireless shares airtime with every device in range, including neighbouring offices and apartments, so performance varies in ways nobody controls. Cable the trading positions and keep wireless for meetings and visitors.
Whether you must is a regulatory question for your compliance advisers. Commercially, disputes are settled by whoever can produce the record, so most firms decide which channels count as business communication, capture those, and make clear that deals are not done on channels that are not captured.
Two controls at different points. Multi-factor authentication on every mailbox prevents the initial compromise. A documented rule requiring telephone verification of any changed bank details, to a number you already hold, stops the payment even when a mailbox has been read. The telephone call is what survives a successful compromise.
It is common and disproportionately risky in a trading firm, because small headcount meets high value. Separate administrator access from daily accounts and hold it with more than one person, keep credentials in a company password manager, document the environment, and have an external provider who already knows the setup.
It should be, and covered hours should match when the desk actually trades rather than an office week. Patching a firewall during a live session is an avoidable self-inflicted outage. Agree the maintenance window against your sessions, including early starts for European and Asian hours.
Connectivity first, since it has the longest lead time and constrains everything else. Then cabling and power protection, which are disruptive to retrofit. Then the security baseline before the desk holds anything valuable. Then the records decision and retention. Backup and monitoring last, with a restore genuinely tested.
Written by Srinivasan Ramachandran, Founder, Listonics · Published Sep 28, 2026 · Last reviewed October 11, 2026 · 2,030 words
