You can use GROUPING SETS for this. The GROUPING function will tell you if a column has been grouped up in that row. It returns 0 for a non-aggregated row, and 1 for a rollup.

SELECT
    YEAR(s.OrderDate) AS Year,
    IIF(GROUPING(MONTH(s.OrderDate)) = 0,
        CAST(MONTH(s.OrderDate) AS varchar(20)),
        'grand_total'
    ) AS Month,
    CASE WHEN GROUPING(MONTH(s.OrderDate) = 0 THEN SUM(sd.UnitPrice) END AS Sum_Price,
    SUM(SUM(sd.UnitPrice)) OVER (PARTITION BY YEAR(OrderDate), GROUPING(Month)
        ORDER BY MONTH(OrderDate) ROWS UNBOUNDED PRECEDING) AS CumSum
FROM
    Sales.SalesOrderDetail sd
JOIN Sales.SalesOrderHeader s ON s.SalesOrderID = sd.SalesOrderID
GROUP BY GROUPING SETS (
    (YEAR(s.OrderDate), MONTH(s.OrderDate)),
    (YEAR(s.OrderDate)
)
ORDER BY
    YEAR(s.OrderDate),
    GROUPING(Month),
    MONTH(s.OrderDate);

You can also use a partial ROLLUP , although it's less clear what's going on. It compiles to the same thing, so efficiency will be the same. You can still use GROUPING() with this method.

GROUP BY
    YEAR(s.OrderDate),
    ROLLUP(MONTH(s.OrderDate))

I don't use this often so this might be off (I tend to let the client code or reporting tool handle the summaries), but try ROLLUP like this:

SELECT
    YEAR(s.OrderDate) AS Year,
    MONTH(s.OrderDate) AS Month,
    SUM(sd.UnitPrice) AS Sum_Price,
    SUM(SUM(sd.UnitPrice)) OVER (PARTITION BY YEAR(OrderDate) ORDER BY MONTH(OrderDate)) AS CumSum
FROM
    Sales.SalesOrderDetail sd
JOIN Sales.SalesOrderHeader s ON s.SalesOrderID = sd.SalesOrderID
GROUP BY
    ROLLUP(MONTH(s.OrderDate)), YEAR(s.OrderDate)
ORDER BY
    Year, Month;