Stock trading

Requirements:

1. Buy and sell financial products via exchange and see their existing positions
2. user should be able to see real time value of their asset and total asset
3. minimise number of server listning to the exchange

NFR: single area high scalable high available low latency

Capacity Estimates:

1. 100M total user
2. 10 % of 100M i.e 10M daily active users for trade
3. 1 user does 5 trade a day
4. 50M trade per day(6 hours a day)= 2K write per sec
5. 30% of 100M i.e (30M * 6*3600) = 6K read
6. 100 different instruments(stock/options ) in portfolio
7. maintain single active connection per client

maintain user connection list to fetch user session connection.

API design stockTicker - NSE/NYCE placeOrder(customerId, stockTicker/exchangeCode, type, quantity, placedAt) fetchStockDetail(String stockId) fetchCustomerPortfolio(customerId) setTrigger(customerId, exchangeCode, stockTicker, quantity, threshold, condition)

DB designs:

Customer: id, name

Stock: 
	id, 
	exchange-BSE/NSE,
	 ticker,
	 price

CustomerStockOrders-
	orderId, 
	customerId, 
	stockId,
	 currentPrice,
	  buy/sell, 
	  status(PLACED/FAILED/STATUS),
	quantity, 
	placedAt
	price
 
Triggers:
	customerId,
	stockId,
	time,
	quantity,
	condition,
	threshold,
	price
	
CustomerPortfolio:
	id
	customerId,
	List<CustomerStock>
	 
	
 
 
StockPriceHistory{
	stockId,
	price,
	timeInmiliSec
}

“

stockExchange

Heigh throughput Exchange systems:

exchange keeps tracks of bids and ask ex:

Bids table:

name	amount	unit
kitty	$100	5
mity	$90	10
Ask Table:

name	amount	quantity
citadel	$120	100
melvin	$130	20
bids will be success full when bidPrice>= askPrice

Matching Machine Service: instead of TCP connection for publising latest pricing and order details, exchange uses UDP multicasting.