Understanding Security Operations Centers
Triage security alerts the way a SOC analyst does, deciding to close, investigate or escalate each one, and explain how a SIEM and a dashboard support that decision.
Who is watching at 2:41 on a Sunday morning
At 2:41 one Sunday morning, someone signs in to Farida Shaikh's staff email from a country Tulsi Health has never done business with. Farida is asleep in Pune, and so is everyone at the company. Yet within fifteen minutes the account is locked and Neha Kulkarni's phone is ringing, because an analyst in Bengaluru saw the sign-in appear on her screen and acted on it. The room that analyst works in is a Security Operations Center, or SOC (said "sock"): a team, with its tools and written procedures, whose job is to watch an organisation's systems around the clock and act quickly on signs of attack.
A control room for computers
Think of the traffic control room of a big city such as Pune. Screens show live feeds from cameras at hundreds of junctions, and a few operators watch them all. They do not drive the cars or repair the roads. When they spot an accident or a jam, they decide how serious it is and call the right people: traffic police, an ambulance, the fire brigade. A SOC does the same for computers. It watches a constant stream of events, meaning anything a computer records as having happened, such as a sign-in, a file being opened or a setting being changed. Each event is written as a line in a log, a computer's diary of what happened and when.
A working SOC needs three things together. It needs people who know what normal looks like and can judge what is not. It needs process, written steps for common situations so that a tired analyst at 3 am handles an alert the same way a fresh one would at noon. And it needs technology that gathers the logs and raises alerts. Tools with nobody watching them are just screens, and people with no tools are guessing.
Staffing a room every hour of every day takes at least five or six people working in shifts, which is far more than Tulsi Health's two-person IT team. So Tulsi Health pays a managed SOC, an outside company that watches many small organisations from one shared room. Tulsi Health's provider is Sentry Desk in Bengaluru. The work is split: Sentry Desk watches and investigates, while Neha decides what the business does about it, because only someone inside Tulsi Health knows which staff member is on leave or which store is closed for a festival.
The managed SOC brings round-the-clock eyes and experience; the in-house lead brings knowledge of the business and the authority to act.
What a SOC analyst does all day
A SOC analyst is a person whose job is to watch, judge and respond to security alerts. Most SOCs arrange analysts in tiers, much as a bank branch works: the counter clerk handles most requests and calls the branch manager only when something is unusual. Tier 1 analysts look at every new alert and decide what it deserves. Tier 2 analysts take the cases passed up to them and dig deeper, tracing what an attacker touched. Tier 3 analysts, the most experienced, hunt for attackers who have not triggered any alert at all and improve the rules that raise alerts.
Every alert enters at the bottom. Only the ones that need more time or skill travel up, so experienced people spend their time where it counts.
Across the tiers the responsibilities are the same few verbs. Analysts watch the queue of alerts, sort them, investigate the doubtful ones, escalate real problems to someone with the authority to act, write down everything they did, and suggest changes to rules that keep crying wolf. The writing down matters more than beginners expect, because the next shift, the customer and sometimes the police may all need to know exactly what was seen and when.
Triage: deciding what each alert deserves
An alert is an automatic message saying that something matched a rule for suspicious behaviour, for example "more than 20 failed sign-ins in five minutes". Alert triage is the quick first sorting of alerts, a word borrowed from hospital emergency wards, where a nurse decides within a minute who must be seen now, who can wait and who only needs a plaster. For each alert the analyst asks three questions: is this real, how bad would it be, and is it still happening? The answer leads to one of three decisions. The analyst can close it as a false positive, an alert that looked bad but turns out to be harmless. They can investigate further when the facts are not yet clear. Or they can escalate, passing it straight to someone who can act, when the danger is clear.
Triage goes wrong in two opposite ways. Closing a real attack as harmless lets the attacker carry on. Escalating every small thing buries Neha in calls until she stops answering, a problem called alert fatigue, where people facing too many alerts begin to ignore all of them. A good analyst checks one or two facts before deciding, and a good SOC keeps tuning its rules so that fewer harmless alerts appear in the first place.
Take the Tier 1 seat at Sentry Desk for one night shift below. Each alert comes from Tulsi Health's systems, and for each one you choose to close, investigate or escalate; a wrong first choice becomes an incident on your shift report.
A SIEM: one notebook for every gate
To see why such a tool is needed, count the places where Tulsi Health keeps logs. The staff email service has one, the customer app and its database (the organised store of customer records) have one, MedLedger on each of the 14 billing laptops keeps one, every store's Wi-Fi router keeps one, and so does the backup. Picture a housing society with four gates, where each guard keeps a separate register. A visitor turned away at Gate 1 who walks round to Gate 3 looks perfectly ordinary in each register. Only someone who reads all four registers side by side notices the same name twice in ten minutes.
A SIEM (Security Information and Event Management, said "sim") is that combined notebook. It collects logs from every source into one place, rewrites them into one common format so that times and names line up, a step called normalising, and then correlates them: it links events from different sources using rules such as "a password reset, a new mail forwarding rule and a large download by the same person within one hour". When a rule matches, it raises one alert instead of three unrelated lines. It also keeps the logs so that analysts can search back weeks later.
A SIEM is only as good as what feeds it. A source that is not connected is a blind spot, and a laptop with the wrong clock puts its events in the wrong place in the story.
A SIEM fails in predictable ways. Any system not sending logs to it is invisible. Computers whose clocks disagree make events appear in the wrong order. And rules that are too eager flood the queue with false positives until the important alert is lost among them.
The log lines below come from four different Tulsi Health sources, merged the way a SIEM shows them. Read them as one story, tick the lines that belong to a suspicious chain of events, and check your choice.
Dashboards: the big screens on the wall
A security dashboard is a screen that summarises what the SIEM sees in numbers, charts and lists, so that an analyst can grasp the state of things in a few seconds. It is the SOC's version of the wall of screens in the traffic control room. A good dashboard answers three questions quickly: is anything serious happening right now, is anything changing compared with a normal day, and what is waiting to be handled?
The counts show how bad things are, the chart shows what changed, and the queue shows what to do next. The red spike and the critical alert are what an analyst's eye should land on first.
Dashboards fail quietly. A dashboard that shows everything green is only as true as the data feeding it: if the Nashik Road laptops stopped sending logs last week, their part of the picture is not "all clear", it is empty. That is why experienced analysts also watch for silence, a source that has stopped reporting, as carefully as they watch for spikes.