Key-value pairs in SQL table - sql

Question

I am building a Django app and I would like to show some statistics on the main page, like total number of transactions, percentage of successful transactions, daily number of active users etc.
I don't want to calculate these values in the view every time a user requests the main page for performance reasons. I thought of 2 possible solutions.
(1) Create a number of one-record tables
Create a table for each of the statistics, e.g.:
from django.db import models
class LastSuccessfulTransactionDate(models.Model):
date = models.DateTimeField()
class TotalTransactionAmount(models.Model):
total_amount = models.DecimalField(max_digits=8, decimal_places=2)
# ...
and make sure that only one record exists in each table.
(2) Create a table with key-value data
class Statistics(models.Model):
key = models.CharField(max_length=100)
value = models.TextField()
and save the data by doing:
from datetime import datetime
from decimal import Decimal
import pickle
statistics = {
'last_successful_transaction_date': datetime(2010, 2, 3),
'total_transaction_amount': Decimal('1234.56'),
}
for k, v in statistics.items():
try:
s = Statistics.objects.get(key=k)
except Statistics.DoesNotExist:
s = Statistics(key=k)
s.value = base64.b64encode(pickle.dumps(v, pickle.HIGHEST_PROTOCOL)).decode()
s.save()
and retrieve by:
for s in Statistics.objects.all():
k = s.key
v = pickle.loads(base64.b64decode(s.value.encode()))
print(k, v)
In both cases the data would be updated every now and then by a cron job (they don't have to be very acurate).
To me solution (2) looks better, because to display the main page I would need to get data from the Statistics table only, not from a number of one-record tables. Is there any recommended solution to this problem? Thanks

Answer 1

There is even better solution if you are using Postgresql, which is JSONField. You can directly store the key value pair in the JSONField. Try like this:
# model
class Statistics(models.Model):
records = JSONField()
# usage
statistics = {
'last_successful_transaction_date': datetime(2010, 2, 3),
'total_transaction_amount': Decimal('1234.56'),
}
Statistics.objects.create(records=statistics)
Statistics.objects.filter(records__last_successful_transaction_date= datetime(2010, 2, 3)) # Running query
s = Statistics.objects.filter(records__last_successful_transaction_date= datetime(2010, 2, 3)).first()
new_statistics = {
'last_successful_transaction_date': datetime(2012, 2, 3),
'total_transaction_amount': Decimal('1234.50'),
}
records = s.records
records.update(new_statistics)
s.records = records
s.save()